Every year, in thousands of Sub-Registrar offices across India, the same quiet transaction takes place under the Registration Act 1908. A seller and a buyer appear before an officer, photographs and fingerprints are taken, a fee and a stamp duty are collected, a document is copied into a register book, and the parties leave with a registered sale deed that both of them believe settles the question of ownership. It does not. The registered deed proves, with the full authority of the state behind the proof, that a particular document was executed by a particular person on a particular day. It proves nothing whatever about whether that person owned the property the document purports to sell. The officer who registered it was forbidden by the design of his own office from asking that question, because the statute that employs him treats the answer as none of his business.

That single fact explains more about Indian civil litigation than any other. The Registration Act 1908, Act 16 of 1908, built a machine for recording transactions in land, and it built that machine well: the machine has run for well over a century, it processes an enormous volume of documents, its records are public, its priorities are clear, and its outputs are admissible in every court in the country. What the machine was never built to do is the thing every buyer asks of it. India registers deeds, not owners. The state maintains a record of transactions and leaves the ownership those transactions assert to be proved, contested, and re-proved in courtrooms, generation after generation. Title in India is therefore presumptive: a registered deed is evidence from which ownership may be inferred, assembled by tracing a chain of documents backwards through decades of transfers, and even the most careful tracing ends in a probability rather than a guarantee, because the state guarantees nothing.

The implementation question this statute poses is unusual, and it is the question this article works through. Most implementation analysis asks why a law fell short of its purpose. This one achieved its purpose with unusual completeness. The registration system does precisely what it was designed to do, which is to create an authentic, dated, publicly searchable record of documents affecting land. The failure that surrounds it, the title searches, the encumbrance certificates, the mutation disputes, the power of attorney sales, the litigation that clogs the civil courts, arises because a century of policy and popular belief has treated that record as though it were a register of ownership, which it is not and never claimed to be. Understanding the Registration Act 1908 therefore requires holding two ideas at once: the machine works, and the machine is not the machine India needs. This article sets out how the registration system is built and what it verifies, what its records prove and what they cannot prove, why stamp duty bends the whole system out of shape, what the Supreme Court did about the power of attorney practice in Suraj Lamp, what the digitisation programmes have and have not changed, and what would genuinely have to be built, in surveys, in adjudication, and in a state indemnity, before India could move from presumptive title to the conclusive, state-guaranteed title that reform documents have promised for decades.

The machine the 1908 Act built

The Registration Act 1908 did not invent deed registration in India, and it did not begin the story. It consolidated a line of earlier statutes, the registration Acts of 1864, 1866, 1871 and 1877, each of which had extended and regularised a system the colonial administration had been building since the early nineteenth century. The 1908 Act remains, in its essentials, the statute those predecessors produced: a procedural code for the presentation, scrutiny, copying and preservation of documents, administered through a hierarchy of registration offices, and organised around the register book rather than around the parcel of land. Its long title describes a statute to consolidate the enactments relating to the registration of documents, and consolidation is the honest word. The Act is about documents. Land enters it only because documents relate to land.

The administrative architecture is a pyramid built for paper. At its head in each state stands an Inspector General of Registration, who superintends the registration establishment across the state. Below the Inspector General sit Registrars of districts, and below them the Sub-Registrars, in whose offices the overwhelming majority of documents are actually presented and registered. The territorial unit of the system is the sub-district: a document relating to immovable property is ordinarily presented in the office within whose sub-district the property, or the part of it governed by the relevant rule in Section 28 of the Act, is situated. The registering officer keeps register books, of which the first, Book 1, holds the copies of documents relating to immovable property, and maintains the indexes prescribed by Sections 54 and 55, an index of the names of parties and an index that allows a property to be traced through the documents that have touched it. Section 52 prescribes the duties of the officer when a document is presented: an endorsement recording the day, hour and place of presentation and the signature of the presenter, a receipt for the document, and, where the document is admitted, its copying into the appropriate book without delay. Section 16A permits those books to be kept in electronic form, a provision whose significance grew as states computerised their registration offices, because it allowed the register book to become a database without any change to the legal character of what the database contains.

Two features of this architecture deserve emphasis, because both shape everything the system can and cannot do. The first is that the register is organised by document and by party, not by parcel. A Torrens register is a folio for a piece of land: one page, one parcel, one owner, one list of encumbrances. The Indian register book is a chronological stream of documents, through which a searcher must reconstruct the history of a parcel by following names and survey numbers across volumes and indexes. The second feature is that the system is a creature of two governments at once. Registration of deeds and documents sits in the Concurrent List of the Constitution, at Entry 6 of List III, which is why a central Act of 1908 governs the subject, while land itself, and the revenue records that describe it, sit in the State List at Entry 18 of List II, which is why every state runs its own revenue record system alongside the registration system, and why stamp duty, the tax that rides on the documents, is substantially a state levy whose rates differ across the country. The fragmentation that buyers experience as three inconsistent records is not an accident of administration. It is written into the constitutional distribution of the subject.

Why did British India build a record of transactions instead of a register of owners?

Because the colonial state wanted two things from land documentation, revenue and notice, and a register of owners would have given it a third thing it did not want, which was responsibility. A deeds registry collected stamp duty and registration fees at every transfer, all without the state ever warranting that a seller owned what he sold.

It created a public record that made transactions harder to deny and ranked competing claims by a visible rule. Guaranteeing title would have required the state first to settle who owned India, a question its own revenue settlements had answered differently in different provinces and often on the basis of possession rather than proof. Recording deeds let the administration monetise and order the market in land while leaving the risk of a bad title exactly where it had always lain, on the buyer.

The deeper history runs through the revenue systems. In the permanently settled tracts of Bengal, in the ryotwari settlements of Madras and Bombay, and in the mahalwari settlements of the north-western provinces, the state’s primary land document was the record prepared for assessing revenue, which identified who was liable to pay and on what holding. Registration of deeds grew up beside that record, in the registration offices created from the 1860s onward, as a service to a growing market in land and to the courts that had to decide disputes within it. The two systems were never merged, because they answered to different departments, different statutes, and different purposes, and the 1908 Act consolidated only the registration half of the pair. Every subsequent reform proposal, down to the conclusive titling drafts of the twenty-first century, has had to confront the same structural fact: India does not have one land record with defects. It has several land records, each internally coherent for its own purpose, which disagree with one another because they were built to answer different questions.

The Act’s own scheme of Parts shows the procedural character plainly. It provides for the registration establishment, for registrable documents, for the time and place of presentation, for the persons who may present documents, for the enquiry the registering officer conducts, for enforcing the appearance of executants and witnesses, for wills and their deposit, for the effects of registration and non-registration, for the duties and powers of registering officers, for fees, and for penalties against improper registration and against the making of false statements and false personation. It is a statute about the integrity of a record. Nowhere in its scheme is there a Part about determining ownership, because determining ownership is the one function the whole design excludes.

What must be registered, what may be, and what the counter actually does

Section 17 of the Registration Act 1908 is the operative heart of the statute, and its categories repay close reading, because they define the perimeter inside which the state’s record is complete and outside which it is not. Section 17(1) makes registration compulsory for instruments of gift of immovable property; for other non-testamentary instruments which purport or operate to create, declare, assign, limit or extinguish, whether in the present or in the future, any right, title or interest, whether vested or contingent, of the value of one hundred rupees and upwards, to or in immovable property; for non-testamentary instruments acknowledging the receipt or payment of consideration on account of such a creation, declaration, assignment, limitation or extinction; for leases of immovable property from year to year, or for any term exceeding one year, or reserving a yearly rent; and for non-testamentary instruments transferring or assigning a decree, an order of a court, or an award, where that decree, order or award itself creates or extinguishes such an interest of the same value. The threshold of one hundred rupees, untouched since the nineteenth century, has long since ceased to filter anything: every real transaction in immovable property exceeds it many times over, so the compulsory net in practice catches every sale deed, mortgage deed, partition deed, settlement deed, release deed and long lease in the country. A proviso lets a state government exempt short leases, those whose terms do not exceed five years and whose annual rents do not exceed fifty rupees, a survival from an era when such figures described real transactions.

Two riders complete the picture of Section 17. Section 17(1A), inserted by the Registration and Other Related Laws (Amendment) Act, 2001, Act 48 of 2001, provides that documents containing contracts to transfer immovable property for consideration, of the kind the doctrine of part performance in Section 53A of the Transfer of Property Act 1882 protects, must be registered if executed after that amendment commenced on 24 September 2001, and that an unregistered such contract has no effect for the purposes of Section 53A. Section 17(2) then carves out a list of exceptions, instruments that the compulsory clauses do not reach, which include among others certain grants and assignments by government, court orders and decrees that are themselves the operative adjudication in defined circumstances, and instruments that merely create a right to obtain another document. Alongside the Transfer of Property Act, the scheme is straightforward to state: Section 54 of that Act provides that a sale of tangible immovable property of a value of one hundred rupees and upwards can be made only by a registered instrument, and the Registration Act supplies the machinery by which instruments become registered.

Section 18 is the permissive counterpart. It lists documents whose registration is optional, a residual category that includes instruments creating or extinguishing interests in immovable property of a value below the Section 17 threshold, leases for a term not exceeding one year, wills, and, at the end, a general clause covering any other document not required by Section 17 to be registered. The will is the instructive entry. A will is a testamentary instrument, Section 17 reaches only non-testamentary instruments, and so the single most consequential document in the transmission of Indian property, the document by which a very large share of holdings pass at death, sits wholly outside the compulsory system. It may be registered, and registration gives it the evidentiary benefits of the register, but the state does not require it, and the register book therefore has a structural silence at exactly the point where most titles change hands.

Time and place discipline the process. Section 23 requires a document, other than a will, to be presented for registration within four months of its execution, with Section 24 adjusting the rule where several persons execute at different times, and Section 25 allowing the Registrar, where delay was caused by urgent necessity or unavoidable accident and the further delay does not exceed four months, to direct acceptance on payment of a fine not exceeding ten times the proper registration fee. Section 27 releases wills from the four-month discipline altogether. Section 28 fixes the venue by the situation of the property. These rules are not technicalities. The four-month window, and the relation-back rule in Section 47 under which a registered document operates from the time it would have commenced had no registration been required, rather than from the date of registration, are what make the register a reliable sequence: the date of execution governs priority, and the deadline forces transactions onto the record while memories and parties are still available.

Presentation itself is regulated by Sections 32 to 35. Section 32 confines the right to present a document to the executant, a person claiming under it, their representative or assign, or an agent authorised by a power of attorney of the kind Section 33 recognises. Section 32A, inserted by the same 2001 amendment Act, requires every person presenting a document to affix a passport-size photograph and fingerprints to it, and, where the document transfers ownership of immovable property, requires the photograph and fingerprints of each buyer and each seller named in the document as well. The provision is the system’s principal defence against personation, the fraud in which a stranger appears as the owner and executes a deed, and its logic is again the logic of the whole statute: the state fortifies the identity of the person at the counter, because identity and execution are what the state certifies. Section 34 then prescribes the enquiry before registration. The registering officer is to enquire whether the document has been executed by the person by whom it purports to have been executed, is to satisfy himself as to the identity of the persons appearing before him and alleging execution, and, where appearance is through a representative, assign or agent, as to the right of that person to appear. Section 35 supplies the outcomes: admission of execution leads to registration, denial leads to refusal, and refusal carries its own appellate path to the Registrar and, beyond that, to a civil suit for a direction to register. Sections 58 to 60 close the sequence: the endorsements on the document admitted to registration, the officer’s signature and date, and the certificate of registration, endorsed on the document, recording that it has been duly registered together with the book, volume and page in which its copy rests.

Read the enquiry in Section 34 once more, because everything in this article follows from its limits. The officer enquires into execution and identity. He does not enquire, and has no statutory warrant to enquire, whether the executant owns the property, whether the property is encumbered, whether a prior transfer exists, whether the description in the deed matches the survey, or whether the consideration stated is the consideration paid. A registering officer who refused a deed because he doubted the seller’s title would be acting outside the Act. The Supreme Court’s treatment of registering officers’ powers runs in the same channel: the office is ministerial in character, its scrutiny is of the document and the parties before it, and the adjudication of title belongs to courts and to no one else in the system. Registration is therefore an act of authentication, not of adjudication, and the certificate at the end of the process certifies registration, not ownership.

What registration proves: the effects provisions

The legal consequences of registration and of its absence are stated in Part X of the Act, Sections 47 to 50, and they are more subtle than the popular account allows. Section 47, as noted, relates the registered document back to its execution. Section 48 provides that registered documents relating to property take effect against oral agreements or declarations relating to the same property, subject to the protection of a transferee who took possession under an oral arrangement that itself constitutes a valid transfer. Section 50 gives a duly registered document of the kinds listed in Section 17(1) and the principal clauses of Section 18 priority over unregistered documents relating to the same property, whether the unregistered document is of the same nature or not, and whether it was executed before or after. The register, in other words, is not merely evidence. It is a ranking system: among competing paper claims to the same land, the registered claim ordinarily wins, which is the practical reason no serious transaction in Indian immovable property is left unregistered where registration is available.

Section 49 is the sanction that makes the ranking system compulsory in fact. It provides that no document required by Section 17, or by any provision of the Transfer of Property Act 1882, to be registered shall affect any immovable property comprised in it, or confer any power to adopt, or be received as evidence of any transaction affecting such property, unless it has been registered. The unregistered sale deed is not a void document in the sense of being a forgery or a nullity between the parties for every purpose; it is a document disabled at the two points that matter. It does not operate on the land, and a court will not receive it as proof that it did. The proviso to Section 49 then keeps three doors open. An unregistered document may be received as evidence of a contract in a suit for specific performance under the Specific Relief Act, as evidence of any collateral transaction not required to be effected by a registered instrument, and, in the text as originally framed, as evidence of part performance for the purposes of Section 53A of the Transfer of Property Act. The third door needs a qualification that many summaries omit: the 2001 amendment Act that inserted Section 17(1A) also removed the part-performance limb from the proviso, so that for contracts executed on or after 24 September 2001 the route to part-performance protection runs through registration itself, and the unregistered agreement is left with the specific performance and collateral uses. The pattern across all three doors is the same. The law will let an unregistered document prove a personal obligation, or a fact collateral to the transfer, but it will not let the document do the one thing registration exists to authenticate, which is to operate as the transfer.

What registration does not do is equally precise, and it is a list of four. Registration does not cure a defect in the executant’s title: a deed executed by a person with no right to the property transfers no right, however impeccable the registration, because a transferor cannot convey what he does not have. Registration does not validate a transaction that another statute forbids or conditions, such as a transfer in breach of a land ceiling condition, a tenancy protection, or a restraint on alienation of tribal land under a state law; the register books record such deeds, and the deeds remain vulnerable. Registration does not reconcile the deed with the other state records: a sale deed describing a plot by an old survey number registers perfectly well even if the revenue record carries the plot under a new number and a different extent. And registration does not extinguish claims that never passed through the system at all: the heir who never mutated, the co-sharer who never partitioned, the holder of an unregistered agreement and possession, the claimant under a will that was never registered and never needed to be. The register is complete only about the documents presented to it. About the land, it is a partial and rebuttable account, and the law has always known this: a registered deed raises a presumption that the transaction it records took place, and shifts the practical burden onto anyone who disputes it, but presumption is the operative word, and presumptions are what courts spend their time rebutting.

Why does a registered document defeat an unregistered one between competing claimants?

Because Parliament chose the register as the ranking device for paper claims to land. Section 50 of the Registration Act 1908 gives a duly registered document of the principal kinds priority over every unregistered document relating to the same property, whether the rival document is older or newer, and Section 47 makes the registered deed operate from its execution date.

The rule rewards the party who placed the claim on the public record and penalises the party who kept a claim private, which is what makes registration close to universal in practice.

What does the certificate of registration handed back at the counter actually certify?

It certifies the fact and the formalities of registration: that the document was presented, that execution was admitted or proved, that the endorsements required by Sections 58 and 59 were made and signed, and that a copy of the document rests in a stated book at a stated volume and page, as Section 60 provides.

It is the state’s attestation of an entry in its register book. It is not a certificate of title, it does not describe the executant as owner, and the Act contains no provision under which it could be read as one.

Section 57 completes the public character of the record, and it is the provision on which the entire practice of title diligence rests. It obliges registering officers to allow inspection of Book 1 and of the indexes relating to it by any person applying, and to give copies of entries in that book, signed and sealed, to all applicants, and it gives those certified copies evidentiary value for the purpose of proving the contents of the original documents. Book 1 is thus a public book in a strong sense: any buyer, lender, journalist or rival claimant may walk into the registration office, or in a computerised state search the database, and reconstruct the registered history of a property. The system’s transparency is real, and it is one of the two genuine achievements of the 1908 design, the other being the priority rule. Everything a buyer can know from the state about a property’s documentary history is knowable because Section 57 made the register book open. Everything a buyer cannot know, the claims that never entered the book, the defects in the claims that did, the state’s silence on ownership, marks the boundary the design drew around itself.

Three records, one plot

The registration record does not stand alone, and its limits become visible only when it is set beside the two other official descriptions of the same land. The revenue record, maintained by the state revenue department through the village and tehsil establishment, is built around the record of rights, the document known by regional names across the country, which records the persons in possession of each holding, the nature of their interest as the revenue administration understands it, the rent or revenue assessed, and the changes effected through mutation as holdings pass by succession, transfer, or partition. The survey record, maintained through the survey and settlement establishment, describes the land geometrically: the cadastral map, the parcel boundaries, the survey numbers and the measured extents. The three records are maintained by different departments, under different statutes, updated by different procedures, and answer different questions. The registration record answers the question, what documents have been executed about this land. The revenue record answers the question, who is treated by the state as holding this land for revenue purposes. The survey record answers the question, where this parcel lies and how large it is. A buyer needs one answer, who owns this land and can sell it to me, and none of the three records gives it.

Record	What it establishes	Who maintains it	What it proves in court	How it can be wrong	What a buyer must therefore check
Registration record (Book 1 and its indexes, under the Registration Act 1908)	That a document was executed by an identified person, presented and registered on a stated date, and copied into the register	The registration department of the state, through Registrars and Sub-Registrars under the Inspector General of Registration	The execution and contents of the document, through certified copies under Section 57; priority over unregistered documents under Section 50	The executant may have had no title; the document may rest on an earlier defective transfer; claims by succession or unregistered dealing never enter it	The chain of registered documents back through successive transfers, read against the identities and descriptions in each deed
Revenue record (record of rights and mutation register)	Who is recorded as holding the land, the character of the holding, and the revenue assessed on it	The state revenue department, through the village revenue officer and the tehsil establishment	A presumption of correctness of the entries, under the state land revenue code, until the contrary is proved; mutation itself confers no title	Entries lag successions and sales by years; mutation may be refused, delayed or made in favour of one heir among several; names and extents drift from the deeds	Whether mutation followed the last transfer, whose names stand recorded, and whether recorded possession matches the seller’s account
Survey record (cadastral map and survey numbers)	The boundaries, shape, survey number and measured extent of each parcel as settled or resurveyed	The survey and settlement department of the state, under the state survey law and settlement operations	The physical identity and extent of the parcel as officially measured	Maps may descend from settlements a century old; subdivisions by sale or inheritance may never have been surveyed; boundaries on paper diverge from boundaries on the ground	The survey number and extent in the deed against the map, and the physical boundaries of the plot against both

The table explains, in one object, why the phrase land records misleads. There is no single record to be accurate or inaccurate. There are three records with three different relationships to the truth, and the litigation begins where they diverge. A deed registers a sale of two acres under an old survey number. The revenue record shows the holding in the name of the seller’s father, mutation after his death never having been carried out, and records an extent of one acre and three quarters. The survey map, drawn at a settlement generations old, shows the parcel undivided, although the family divided it among four sons decades ago by an unregistered arrangement the whole village recognises. Every document in that story is genuine. Every record is doing its job. And the buyer who relies on any one of them will buy a lawsuit with the land.

Mutation deserves a paragraph of its own, because the confusion between mutation and title is the revenue record’s equivalent of the confusion between registration and title. Mutation is the updating of the revenue record to reflect a change: a death, a sale, a gift, a decree. It is a fiscal and administrative act, performed so that the state knows from whom to demand revenue and to whom to address its notices. The Supreme Court has stated the consequence in terms that admit no qualification, in Sawarni v Inder Kaur, (1996) 6 SCC 223, and repeatedly since: mutation of a property in the revenue record does not create or extinguish title, and the entries have no presumptive value on the question of ownership; they are relevant for the fiscal purpose for which they are made. The recorded holder may be a trustee, a benamidar, a widow holding for the family, or a stranger whose application no one opposed. Yet in village India the record of rights is the document people actually hold, cite, and fight over, because it is the only one of the three records that names a person against a parcel in a single line. The registration record names parties to documents. The revenue record names holders of land. The psychological pull of the second over the first is understandable, and it is exactly backwards in law: of the two, the revenue entry is the weaker evidence of title, and the registered chain is the stronger, and neither is conclusive.

Stamp duty: the fiscal engine of the distortion

No account of the registration system’s performance is honest unless it begins with the tax. A deed becomes registrable only when it is duly stamped under the Indian Stamp Act 1899 or the corresponding state law, and stamp duty on conveyances is levied ad valorem, as a percentage of the consideration or of the market value of the property, at rates set by states. Registration fees ride on top. The combined impost on a transfer is therefore a substantial fraction of the price, varying by state and by the character of the parties and the property, and it is payable in cash, at the moment of transfer, by parties who frequently experience it as the single largest transaction cost of the purchase. Every pathology of Indian conveyancing flows from that arithmetic. A tax of that size, levied on a stated figure that the parties themselves supply, creates an overwhelming incentive to state a smaller figure.

The first consequence is undervaluation. The deed recites a consideration below the price actually paid, the balance moves outside the document, and the state responds with machinery of its own: circle rates, guideline values, and ready reckoner rates that fix a floor below which the registering authority will not accept the stated value for duty purposes, supported by provisions such as Section 50C of the Income-tax Act 1961, under which the stamp duty value is substituted for the stated consideration in computing capital gains where it exceeds that consideration, a point at which the registration record and the income tax record are made to check each other. The floor limits the loss without removing the incentive, and the recorded consideration in a large share of Indian deeds remains an artefact of duty management rather than a statement of price. The register book, faithfully recording the document as executed, faithfully records the fiction as well, so the state’s own record of the land market is systematically corrupted at the point of price, a corruption no digitisation of the record can repair because the falsehood is in the document, not in its storage.

The second consequence is avoidance of the registered deed altogether, and its name is the power of attorney sale. Where duty on a conveyance is heavy, parties learned to achieve the economic substance of a sale while executing no conveyance at all. The seller executes an agreement to sell, which in many states attracted a lower duty when unaccompanied by possession, an irrevocable general power of attorney authorising the buyer or the buyer’s nominee to deal with the property, and a will bequeathing the property to the buyer, and the buyer pays the full price and takes possession. No sale deed is presented, no conveyance duty is paid, and the chain of title acquires a link that is not a link at all. The practice flourished in the metros, above all in Delhi, where it was nourished by a second distortion as well: restrictions on the transfer of leasehold and other regulated tenures made the power of attorney route a way around consent requirements and conversion charges, and an entire market of property dealers normalised what lawyers called GPA sales. The registration system’s perimeter in Section 17 was not breached by force. It was walked around, by transactions deliberately structured to present nothing the section could catch, at a fiscal saving that buyers and sellers divided between them.

The third consequence is slower and larger. Because duty is a state subject in its rates and a politically visible tax, states compete and hesitate: a state that cuts duty sharply loses revenue on every transaction in the hope of gaining volume and honesty, a state that holds rates high protects revenue and manufactures evasion. The Supreme Court itself, in the judgment examined next, urged states towards lower duty as the structural remedy, observing that high duty was the engine of undervaluation and of the black money that moves through property. The fiscal design and the title system are therefore not two subjects. The quality of India’s land record is set, in significant part, in state finance departments, by officers deciding a percentage, and every proposal for conclusive titling that ignores the duty structure proposes to build a guaranteed register on top of a transaction stream that the duty structure has already taught to lie.

Suraj Lamp and the limits of a judgment

The judicial anchor of the modern law is Suraj Lamp and Industries Private Limited v State of Haryana, (2012) 1 SCC 656, decided on 11 October 2011. The case reached the Supreme Court through a special leave petition arising from the Haryana development authority’s refusal to act on power of attorney transactions, and the Court, speaking through Justice R. V. Raveendran, used it to state the law on an entire practice. The holding is narrow and absolute. Transactions of the nature of general power of attorney sales, or sale agreement, general power of attorney and will transfers, do not convey title and do not amount to transfer, nor can they be recognised as a valid mode of transfer of immovable property. The Court worked through each instrument. An agreement to sell, under Section 54 of the Transfer of Property Act, creates no interest in or charge on the property; it is a contract, enforceable by a suit for specific performance, and nothing more. A power of attorney is an instrument of agency under the Indian Contract Act 1872; the attorney acts for the principal, acquires no ownership, and the description of the power as irrevocable does not convert agency into conveyance. A will speaks only from death, is revocable during the testator’s lifetime, and cannot function as an inter vivos transfer. Only a deed of conveyance, duly stamped and registered, transfers title.

The reasoning matters as much as the holding for an implementation analysis, because the Court located the practice in the fiscal and administrative system rather than in the morality of the parties. It recorded that such transactions were resorted to in order to avoid the payment of stamp duty and registration charges, that they fed the circulation of unaccounted money, that they cheated the revenue, and that they generated enormous litigation and criminality, including the seizure of properties by force and the exploitation of purchasers who discovered that the person who had taken their money had nothing to sell or had sold it several times over. It directed that such transactions not be treated as completed transfers by authorities, and that municipal and revenue bodies not effect mutation on their basis. And it said, expressly, that the remedy lay partly with the states, in reducing stamp duty so that the honest route stopped being the expensive one.

The judgment’s practical effect was substantial and incomplete, and both halves of that verdict are the implementation story. Substantial, because after October 2011 no authority could treat a GPA sale as a transfer, banks and housing finance companies tightened their insistence on registered conveyance, the conveyancing profession acquired an unimpeachable citation for advice it had long given, and the practice lost its semi-respectable standing in the organised market. Incomplete, because a judgment can declare law but cannot repeal arithmetic. Where the duty saving remains large, where a tenure cannot lawfully be transferred without charges the parties wish to avoid, where a seller’s title is itself a GPA link that no registered deed can be built upon without first regularising the chain, the practice continues in the informal market, priced as a discount for risk. The buyers in that market are typically those least able to bear the loss when the title fails. Suraj Lamp thus stands as the cleanest available demonstration of the series thesis this cluster keeps meeting: a rule perfectly stated, by the highest court, in the clearest terms, changes behaviour only where the surrounding incentives let it. The registration system did not fail to absorb Suraj Lamp. The judgment exposed that the system’s perimeter is set by fiscal policy, and fiscal policy was not before the Court.

The litigation figure belongs here, handled with the discipline the subject requires. It is frequently said that a large majority of Indian civil cases concern land, and the figure most often quoted, of about two-thirds, traces to a specific source: the Access to Justice Survey conducted by DAKSH, a Bengaluru-based organisation working on judicial reform, which interviewed more than nine thousand litigants in three hundred district courts across twenty-four states between November 2015 and February 2016, and found that 66.2 per cent of the civil cases in its sample were land or property disputes. That is a survey of litigants in subordinate courts, not a census of filings, and its figure should be quoted with its method attached. Other counting methods produce other numbers: a study of the Delhi High Court’s decided cases by the researchers Damle and Gulati, characterising land and property related litigation at that court, classified about seventeen per cent of its case set as immovable property related, on a case set that included criminal matters and disputes to which no private citizen was a party, in a single metropolitan High Court. The two figures do not contradict each other; they count different things. What every method agrees on is the direction: land and property disputes form the largest single family of civil litigation in India, land disputes take years and commonly decades to resolve, and a registration and records system that cannot answer the ownership question guarantees that the courts will be asked it instead.

The digitisation record: computerisation without conversion

The modernisation programme is the implementation record proper, and it has a longer history than its name suggests. The Union government began funding the computerisation of land records in 1988-89 through the Computerisation of Land Records scheme, and the strengthening of revenue administration and updating of land records through a companion scheme begun in 1987-88. Both ran for two decades as assistance to states, with uneven results, before the Cabinet approved their merger and expansion on 21 August 2008 into the National Land Records Modernization Programme, whose stated ambition reached beyond computerisation to a modernised, integrated land record system and, in its programme documents, to an eventual system of conclusive titles. On 1 April 2016 the programme was revamped and renamed the Digital India Land Records Modernization Programme, converted into a central sector scheme with one hundred per cent central funding, and organised around components that map exactly onto the three-records problem: computerisation of the record of rights, digitisation of cadastral maps, computerisation of registration, and integration of the registration and revenue systems so that a registered transfer triggers the mutation process rather than waiting for a separate application. The Department of Land Resources in the Ministry of Rural Development administers the programme, developed a National Generic Document Registration System as common software for registration offices, and built the Unique Land Parcel Identification Number, a fourteen-digit alphanumeric identifier generated from the geo-coordinates of the parcel, described by the Department as an Aadhaar for land.

The output figures, stated with their sources and dates because the programme’s statistics are meaningless without them, describe a genuine transformation of the record’s medium. A release of the Press Information Bureau dated 1 January 2026 reported that computerisation of the record of rights stood completed in 97.27 per cent of villages, that cadastral maps had been digitised for 97.14 per cent of the country, that the textual record of rights and the spatial cadastral map had been integrated in 84.89 per cent of villages, and that citizens in nineteen states could download digitally signed, legally valid land records without visiting an office. The Minister of State for Rural Development, Dr Chandra Sekhar Pemmasani, answering a question in the Lok Sabha with figures drawn from the programme’s management information system as on 9 February 2026, reported 99.88 per cent of records of rights computerised, 97.41 per cent of cadastral maps digitised, and 96.61 per cent of Sub-Registrar offices computerised, against figures of 91 per cent, 68 per cent and 93 per cent respectively five years earlier, and reported that more than thirty-eight crore land parcels had been assigned the Unique Land Parcel Identification Number. The two sets of figures measure overlapping things on different denominators, villages against records, and should not be blended; what they jointly establish is that the scanning, data entry and office computerisation components of the programme are near saturation, and that the integration components lag the digitisation components by a margin of roughly fifteen percentage points on the village measure.

The weaker components of the programme repay attention because they show where the physical system resists. The Parliamentary Standing Committee on Rural Development and Panchayati Raj, reviewing the programme in a report presented in 2026, recorded the digitisation components at the saturation levels the ministry reported and placed the modern record rooms, the fireproof and indexed physical archives in which the original register books and deed volumes rest, at 72.5 per cent, with the integration of the textual and spatial records in 84.32 per cent of villages as on 31 January 2026. The register book is not superseded by its scan: the certified copy a court acts on derives its authority from the book and its electronic successor record together, and the preservation of the paper archive remains a live component of the digital programme rather than a leftover from the paper one. A programme audit that reads only the percentages misses this layering. Each phase of the modernisation has added a stratum, paper book, scanned image, database entry, geo-referenced parcel, without retiring the strata beneath, and the citizen’s title is distributed across all of them.

The urban record, the weakest of the three, acquired its own instrument in the NAKSHA pilot, the National Geospatial Knowledge-based Land Survey of Urban Habitations, which the Press Information Bureau release of 1 January 2026 reported as progressing across 157 urban local bodies, using aerial survey to build the urban land record that municipal government never possessed. The need is structural. Urban India keeps its property account in municipal tax registers compiled for assessment, a fourth record beside the three already described, and the constitutional design of municipal government under the 74th Amendment gave urban local bodies functions without giving them a land record of their own, so that the most valuable land in the country is covered by the least systematic record. A buyer of a city flat assembles title from the developer’s chain, the housing society’s conveyance, the municipal assessment, and the registration indexes, and the absence of a surveyed urban parcel record is one reason the Rajasthan experiment, examined below, confined itself to urban land where values justified the effort.

The analytical point follows, and the programme’s own reporting supports it. Digitisation changed the medium of every record it touched and the legal character of none of them. A computerised record of rights is still a fiscal record with a presumption of correctness and no title in it. A digitised cadastral map is a faithful copy of a map that may descend from a settlement conducted generations ago, its errors preserved at higher resolution. A computerised registration office produces the same certificate of registration, attesting execution and entry, that the paper office produced. Integration, the component that lags, is the one that attacks a real defect, the divergence between the registration and revenue records, because an automatic mutation trigger removes the years-long gap in which a registered sale leaves the revenue record standing in the seller’s name. But integration harmonises records with each other; it does not harmonise them with the truth. The programme was conceived, in 2008, as a road to conclusive titling. Measured against that destination, its achievement is better described as the complete modernisation of the presumptive system: the presumption is searchable online, downloadable in nineteen states, attached to a fourteen-digit parcel number, and rebuttable in exactly the same courts as before.

What changes when a paper land record becomes a digital one, and what stays exactly the same?

The medium changes completely: the record becomes searchable from anywhere, copies issue without the mediation of the village accountant, tampering leaves traces, and a registration can trigger mutation automatically where the two systems are integrated. The legal character does not change at all.

A digital record of rights remains a fiscal record, a digital register book remains a record of documents, and neither becomes evidence of a state-guaranteed ownership that the underlying law never created.

The transactions the register never sees

The register’s silence has a structure, and mapping it explains why even a flawless Section 57 search leaves risk behind. The largest silence is succession. A very large share of Indian land passes at death, and the instruments of that passage sit at the edge of the system or outside it. A will, as shown, is registrable only at the testator’s choice. Intestate succession requires no instrument at all: title devolves by operation of the personal law the moment death occurs, and the only public trace is mutation, which the heirs may seek years later or not at all. Probate, the court’s certification of a will, is compulsory only within the original civil jurisdiction of the High Courts at Bombay, Madras and Calcutta for the wills to which Section 213 of the Indian Succession Act 1925 applies, a territorial survival that leaves most Indian wills unprobated as well as unregistered. The consequence is a class of titles that are legally complete and documentary invisible: the seller is owner by inheritance, his name appears in no deed, and the buyer’s diligence must reconstruct the family, the deaths, the marriages and the relinquishments that the register was never asked to hold. Fraud and litigation concentrate here precisely because the record is thinnest here.

The second silence is the mortgage by deposit of title deeds. Section 58(f) of the Transfer of Property Act 1882 permits a borrower in the presidency towns and other notified towns to create a mortgage by delivering the title deeds to the lender with intent to create a security, without any instrument and therefore without registration. The device was built for commercial speed, and it remains the standard form of housing and business security in the towns where it operates. A transaction of the first importance for title, an encumbrance that binds the land against later buyers who fail to ask why the seller cannot produce his original deeds, is thus created by a private act of delivery that the register cannot record. The state’s answer, built a century later, is a second register: the Central Registry of Securitisation Asset Reconstruction and Security Interest of India, CERSAI, established under the SARFAESI Act 2002 and operational from 2011, in which security interests including equitable mortgages are filed, so that a lender’s diligence runs through two searches, the Sub-Registrar’s indexes for deeds and the central registry for charges, each incomplete in a different way. The pattern is characteristic of the whole field. Every gap in the 1908 record has been answered not by widening the record but by adding another one beside it.

The third silence is possession without paper. Agricultural holdings are divided among sons by arrangements the family honours and no office records; tenants and sharecroppers cultivate under understandings that tenancy legislation in many states discourages them from writing down; occupants hold under agreements to sell that were stamped inadequately or not at all and never presented; and adverse possession matures in the background of neglected holdings, converting long occupation into a plea against the documentary owner. None of this appears in Book 1. The register is an accurate record of the formal economy of land laid over an informal economy of land that operates beneath it, and the friction between the two economies is a large part of what the civil courts process. This is also the deepest reason the conclusive titling project cannot be reduced to a surveying and computing exercise: a register of titles opened tomorrow would have to decide, parcel by parcel, how much of the informal economy to recognise, and every answer creates a class of losers with a grievance against the state rather than against each other.

How the courts work with the record the system gives them

Litigation is where the presumptive character of the record is cashed out, and the courts have developed a settled practice for working with registered deeds that any implementation account should describe, because it defines what registration is worth in the forum where title is actually decided. A registered deed is strong evidence of its own execution. The endorsements and certificate under Sections 58 to 60 are themselves evidence of the facts they record, and a party who denies execution of a registered document faces the practical burden of displacing an official record made at the time, in the presence of the parties, by an officer with no interest in the outcome. Where the document is of a kind the law requires to be attested, a gift deed or a mortgage deed, the evidence statute requires an attesting witness to be called to prove execution while one survives and is available; a sale deed, which the law does not require to be attested, is proved by its registration record and the testimony of those present. Registration, in short, manufactures the best evidence of the transaction that the legal system knows how to make, which is why forgery attacks on registered deeds so often take the form of attacking personation at the counter, the one point Section 32A was written to defend, rather than attacking the record itself.

The converse practice governs the unregistered document, and Section 49’s proviso has generated a jurisprudence of collateral use that reveals the courts’ sympathy pulling against the statute’s discipline. An unregistered lease deed cannot prove the lease, but courts have received such documents to prove the character of the tenant’s possession, because the nature of possession is a fact collateral to the transaction the document attempted. An unregistered partition deed cannot effect the partition, but it may evidence a severance of joint status that the parties’ conduct independently shows. The line the courts patrol is the line Section 49 draws: the document may prove facts around the transaction, never the transaction’s operation on the land. The doctrine of part performance in Section 53A of the Transfer of Property Act was the great historical exception, protecting a transferee in possession under a written contract who had performed his part, and the 2001 amendments to both statutes, inserting Section 17(1A) and recasting the Section 49 proviso, deliberately narrowed it, requiring registration of the very contracts the doctrine protected. The legislative direction of travel is unmistakable across a century: the perimeter of the compulsory record keeps expanding, the tolerated informalities keep shrinking, and the system’s centre of gravity moves steadily from possession towards paper. What does not move is the boundary at the far side of registration. However complete the paper becomes, the courts still decide ownership behind it, deed by deed and presumption by presumption, because no statute has authorised anyone else to.

Refusal and its remedies: the Act’s own appeal ladder

The counter at which registration happens has a working life that the statute regulates more closely than its reputation suggests, and the refusal provisions are where that regulation bites. A Sub-Registrar who refuses to register a document must record the reasons in Book 2 under Section 71 and give a copy of the reasons to the person executing or claiming under the document, without unnecessary delay, on application. From that refusal the Act builds a short ladder with short deadlines. Section 72 gives an appeal to the Registrar within thirty days of the order, the Registrar being empowered to reverse the refusal and direct registration. Where the refusal rests on denial of execution, the route is different: Section 73 gives the person claiming under the document thirty days to apply to the Registrar, who then enquires under Sections 74 and 75, summoning and enforcing the attendance of witnesses with the powers of a civil court for that purpose, and orders registration if execution and the requirements of the Act are established. The Registrar’s own refusal is recorded with reasons under Section 76, no appeal lies from it, and Section 77 closes the ladder with a suit: within thirty days of the Registrar’s order, the claimant may sue in the civil court within whose jurisdiction the office sits for a decree directing registration, and the document, if the decree passes, is presented again within thirty days of it.

Three features of this ladder are implementation evidence. The deadlines are thirty days at every rung, which tells the citizen that the system expects refusals to be cured promptly and penalises the party who treats refusal as a resting state. The reasons must be recorded in a numbered book and supplied on application, which makes refusal a documented act exposed to the ladder above it rather than a discretionary shrug at the counter. And the Registrar’s enquiry is a genuine adjudication, with witnesses and costs, confined to the single question the whole system is allowed to ask: was this document executed by the person named, and have the Act’s requirements been met. Even at the top of its own appellate structure, the registration system does not adjudicate title. It adjudicates execution, one document at a time, and the discipline never relaxes.

Around the counter stands the profession the system created. Most deeds in India are not drafted by the parties or, in the smaller markets, by advocates; they are prepared by deed writers, licensed under the registration rules of the states, who cluster around the registration offices, hold the precedents, know the Sub-Registrar’s requirements, compute the duty, and shepherd the document through presentation. The institution is a rational response to a procedural statute: a fixed-form transaction, repeated daily, generates a specialist intermediary class, and the quality of the record depends in part on the quality of that class, since the description of the property, the recital of the chain, and the statement of consideration all pass through the deed writer’s draft before any officer sees them. The standardisation that computerised systems brought, pre-approved formats, online duty calculators, token systems for presentation, narrowed the deed writer’s discretion without removing the function, and the online registration that the draft replacement statute proposes would move the function rather than abolish it, from the office steps to the preparer’s desk. An implementation account that ignores this layer misdescribes the system as citizens experience it: for most parties, the registration system is the deed writer, the stamp vendor, and the counter, in that order, and the officer whose enquiry the statute prescribes is the last and briefest of the three encounters.

The fee table and the penalty provisions: the system’s fiscal and criminal skeleton

Two Parts of the Act that commentary rarely reads complete the implementation picture, because they state what the system charges and what it punishes. Part XIII, on fees, gives the fee structure to the states: Section 78 requires the state government to prepare the table of fees payable for registration, for searching the registers, for copies, and for the ancillary services of the registration office, Section 79 requires the table to be published in the Official Gazette and displayed in every registration office, and Section 80 makes the fees payable on presentation. The sums look modest beside the stamp duty, and for the citizen they merge into a single payment at the counter. Fiscally, however, the combined flow of stamp duty and registration fees is one of the largest heads of own revenue available to state governments, collected at a point the citizen cannot avoid, by an establishment whose costs are a small fraction of its yield. That fiscal fact is a standing part of the explanation for the system’s durability and its distortions alike: the registration office is a revenue office, staffed and supervised as one, and its success is measured in collections long before it is measured in record quality. A titling guarantee would invert the office’s economics, turning a profit centre into a liability holder, and the reluctance of state finance departments to volunteer for that inversion is a rational obstacle that reform documents rarely budget for.

Part XIV, on penalties, draws the criminal perimeter around the counter. Section 81 punishes a registering officer, or a person employed in the registration office, who endorses, copies, translates or registers a document in a manner he knows or believes to be incorrect, intending thereby to cause injury, with imprisonment that may extend to seven years, or fine, or both. Section 82 punishes the user side of the counter with the same maximum: intentionally making a false statement before an officer acting under the Act, intentionally delivering a false copy or translation or a false copy of a map or plan, falsely personating another in any proceeding under the Act, and abetting any of these. Read the two sections together and the statute’s theory of fraud is laid bare. The frauds it criminalises are frauds against the record: the corrupted entry, the false statement in the enquiry, the impersonated executant. The fraud it does not criminalise, because it cannot, is the honest registration of a deed by a man who is exactly who he claims to be and owns nothing of what he sells. That fraud belongs to the general law of cheating and forgery, and its victim’s remedy lies in the civil courts, because the registration system has authenticated everything about the transaction except the one thing the victim needed. The penalty provisions thus close the design’s logic with unusual candour. The state polices the integrity of its book with seven-year sentences. The truth of what the book records about ownership, it leaves, as it has always left it, to the parties and the courts.

Where the machine works

An implementation verdict that lists only failures would misread this statute, because the registration system performs real work that a title system would also have to perform, and performs some of it better than its critics allow. It deters the crudest frauds: the requirement of personal appearance, the photograph and fingerprint discipline of Section 32A, the public character of Book 1, and the criminal provisions of the Act against false personation and false statements raise the cost of the forged deed and the impersonated seller, historically the commonest engines of property fraud, to a level at which they require organisation rather than opportunity. It supplies the priority rule that lets a market function at all: because Section 50 ranks registered over unregistered claims and Section 47 relates registration back to execution, a buyer who registers promptly acquires a defensible position against later claimants through the same paper, and lenders can take mortgages with a calculable risk. It creates the raw material of diligence: Section 57’s open register is why the Indian title search exists as a professional service, and why institutional credit against urban property, whatever its frictions, is possible at national scale. And it raises, through stamp duty and fees, a very large revenue for states, which is not a trivial entry in the ledger of a system whose fiscal yield is one reason reform is politically expensive.

The comparison the implementation framework demands, against a similar system that implemented differently, is available inside the same legal order. The revenue record system and the registration system are both century-old state records of land, both computerised under the modernisation programme, and their performances have diverged in an instructive direction. The registration record, for all its limits, is a record of acts performed before an officer at a stated time: its entries are hard to falsify retrospectively, its sequence is disciplined by the four-month rule and the relation-back doctrine, and its public inspection is a statutory right. The revenue record, by contrast, depends on continuous updating by the lowest tier of the revenue establishment in response to events, deaths, partitions, informal divisions, that nobody is obliged to report and many have reasons to postpone, and its entries decay from the day they are made. The lesson generalises beyond land. Records of transactions, refreshed by the parties’ own interest at the moment of the transaction, stay accurate. Records of status, refreshed only by administration, rot. A conclusive title register is a record of status, the hardest kind to keep true, and any state proposing to maintain one should first explain how it will outperform the mutation register at the very task the mutation register fails.

The acquisition system shows the cost of the divergence from the other side. When the state acquires land, it must identify and compensate the persons interested in it, and it works from the revenue record, the record least able to bear the weight: the result, repeated across projects, is compensation paid to recorded holders while unrecorded claimants litigate for decades, acquisition awards reopened by rival title claims, and projects delayed by disputes the acquiring authority had no instrument to foresee, a dynamic the land acquisition framework must manage at every notification. Unclear title is not only a private misfortune between buyer and seller. It is a tax on every public work that touches land.

The honest verdict on the 1908 Act as implemented is therefore a verdict about a boundary. Inside its design perimeter, the system is a success: documents are authenticated, sequenced, preserved, opened to inspection, and ranked, at a scale of hundreds of thousands of registrations a year in the larger states, by an establishment whose basic procedures a citizen can learn and use. At the boundary, the system stops, by design, exactly where the citizen’s need begins: it will not say who owns the land. The persistence with which buyers, lenders, journalists and even governments speak of registered title, clear title from the registrar, and title verified by registration shows a demand for a state guarantee so strong that it projects the guarantee onto a record that never offered it. A statute cannot be faulted for failing to do what it forbade itself to attempt. The policy failure of the century is different: knowing the boundary, successive governments treated the deeds record as if it discharged the state’s responsibility for the ownership question, and funded the guarantee’s prerequisites, survey, adjudication, indemnity, last and least.

One central Act, fifty registration systems

Because registration sits in the Concurrent List, the 1908 Act is a central statute that states may amend for their own territories with the President’s assent, and a century of state amendments has produced material variation inside a single statutory frame. States have widened the compulsory list in Section 17 for their territories, several of them moving to require registration of agreements to sell and of powers of attorney connected with immovable property transfers, tightening the perimeter that the power of attorney practice exploited. States have rebuilt the procedure: Maharashtra’s system of e-registration for leave and licence agreements, long the standard for urban renting in that state, dispenses with the physical counter visit for an entire class of documents, and several states have built online search and certified copy services on top of the computerised register that go well beyond what the parent Act contemplated. Stamp duty, whose rates for conveyances are a state subject, varies widely enough that the same transaction carries a markedly different fiscal load on opposite sides of a state border, and registration fees, exemptions for classes of transferees, and concessions, for women purchasers in a number of states among them, are all set state by state. The Inspector General in each state heads what is in substance a state registration service administering a central text.

For the implementation question, the variation is evidence rather than noise. It shows that the operating constraints on the 1908 system are not statutory. The states that computerised earliest and integrated registration with mutation did so under the unamended central scheme, using the electronic-books provision and their rule-making powers; the states whose records remain least reliable are not distinguished by a different Registration Act but by the capacity of the same establishment. It also shows where reform energy actually goes when it is left to the level of government that owns the problem: towards duty rates, towards perimeter widening, towards counters and databases, and away from the ownership question, which no state except Rajasthan has touched. Federalism in this field is not the obstacle to a title system. It is the reason a title system, if it comes, will arrive state by state at different speeds, which is precisely how the Rajasthan precedent is built.

The economics of diligence: two markets for Indian land

The presumptive system sorts the land market into two tiers, and the sorting is one of its least discussed implementation effects. In the institutional tier, developers, listed companies, banks and housing finance companies buy certainty as a service. A project begins with a title investigation by counsel, a report that traces the chain, certifies the searches, lists the encumbrances and opines on marketability; the Real Estate (Regulation and Development) Act 2016 then requires the promoter to place the title position before the public, disclosing the legal title report and the encumbrance position for the land on which development is proposed, so that the diligence for a regulated project is performed once, professionally, and published. Lenders in this tier search the registration indexes and the CERSAI registry, take registered mortgages or documented equitable mortgages, and price the residual risk into the loan. Transactions in this tier are slower and more expensive than the parties want, and they close.

In the individual tier, a family buying a plot or a smallholding buys diligence retail or does without it. The search fees are small; the professional fee for a genuine chain investigation is not, relative to the transaction, and the temptation to rely on the seller’s papers, the revenue entry, and local knowledge is correspondingly strong. The information asymmetry is structural: the seller knows the family’s succession history and the buyer cannot, the broker is paid on completion and not on accuracy, and the state’s records, as the three-records analysis showed, answer every question except the one being asked. The losses in this tier are borne by households least able to absorb them, and they surface in the DAKSH survey’s finding, drawn from litigants across every income group, that land and property disputes dominate the civil docket of the district courts. A registration system is often evaluated as a piece of legal machinery. It should also be evaluated as a distributive institution: presumptive title is a subsidy to those who can buy diligence and a tax on those who cannot, and the case for a state guarantee rests at bottom on that distributive fact rather than on administrative tidiness.

The case for conclusive titling, at full strength

The alternative design is neither new nor exotic. Title registration of the Torrens kind takes its name from Sir Robert Richard Torrens, premier of South Australia, whose Real Property Act of 1858 replaced deeds registration in that colony with a register of titles: the state examines a claim once, enters the owner on a folio for the parcel, and thereafter the register is the title, conclusive against the world, with transfers effected by entry on the folio and a state assurance fund compensating anyone injured by the register’s operation. The system’s three working principles are conventionally summarised as the mirror, the curtain, and the insurance: the register mirrors the title completely, so a buyer need look nowhere else; the curtain keeps the history and the trusts behind the register out of the buyer’s inquiry; and insurance, through the assurance fund, answers for the cases where the mirror is wrong. Jurisdictions across the Commonwealth adopted versions of the model, and England built its own registered title system by statute over the following century.

The case for India, made in its strongest form, runs as follows. The Peruvian economist Hernando de Soto, in The Mystery of Capital, published in 2000, argued that the poor of the developing world hold their assets in defective forms, as houses on unrecorded land and holdings without formal title, which cannot be mortgaged, cannot secure credit, and cannot be traded beyond a circle of personal knowledge: dead capital, in his phrase, wealth that exists physically and not economically. India’s presumptive system manufactures dead capital at scale. Every transaction requires the buyer to purchase the same diligence again, a search through decades of deeds, because the last buyer’s diligence created no public good: it settled nothing for the next buyer. Credit against land is rationed by the lender’s fear of the unknown claimant, and the fear is rational. Litigation consumes the civil courts. A conclusive register converts diligence from a private cost repeated at every transfer into a public investment made once, replaces a chain of inference with a state warranty, and gives the dispossessed claimant a defendant who can pay, the state and its fund, instead of a seller who may be dead, absent, or judgment-proof. On this account, presumptive title is not a neutral inheritance; it is a continuing decision to privatise the cost of uncertainty onto the parties least able to bear it, and to reserve the benefits of certainty to those who can buy diligence.

There is force in every element of that case, and the article’s verdict does not deny it. The obstacles belong to a different order of argument: not whether a conclusive register would be better than the presumptive system if it existed and were accurate, but what building one on the Indian record base would actually do.

The obstacles the advocacy understates

A conclusive register is only as good as its first entry, and the first entry is where the Indian difficulty concentrates. To open a folio for a parcel, the state must decide who owns it, conclusively. That decision requires, first, a settled survey: boundaries fixed on the ground and on the map, subdivisions recorded, parcels identified. The survey record, as the three-records analysis showed, descends in many districts from settlements of great age, resurvey has proceeded unevenly, and the digitisation programme’s own figures, with geo-referencing of cadastral maps standing well below the digitisation figures in the programme reporting, measure a spatial base that is not yet settled. It requires, second, the resolution of every existing competing claim: the unmutated successions, the unregistered partitions, the possessions that ripened or failed to ripen into rights, the tenancies and customary interests that no deed ever recorded. In a deeds system those claims sleep until a transaction wakes them. In a titling operation they must all be adjudicated at once, by the state, in a single administrative campaign, with the losers extinguished or compensated. It requires, third, an indemnity fund with real money in it: a state that guarantees titles will be liable when the register is wrong, and the guarantee is credible only if the fund can pay claims that will be measured against urban land values. India, as the brief for reform itself usually concedes in its later pages, has none of the three at scale: no completed modern survey, no machinery that has ever attempted a universal first adjudication of claims, and no assurance fund in any state.

The consequence of building the register anyway is the point the counter-reading insists on, and it should be stated plainly. A conclusive register erected on unreliable underlying records does not remove uncertainty. It converts uncertainty into wrongful certainty, and wrongful certainty is worse, because it is armed. Under the presumptive system, a weak claim can be tested in court before it destroys a transaction; under a conclusive system built on a defective first registration, the register itself destroys the true owner’s claim, leaving compensation, if the fund pays, in place of the land. The risk falls predictably: on holders whose rights are real but undocumented, on heirs who never mutated, on occupants under customary and informal tenure, on women whose interests in family land are recognised in practice and recorded nowhere. A title campaign that demands documentary proof from a population whose holdings predate documentation will not merely make errors; it will make them in one direction, in favour of the documented against the undocumented, and it will give those errors the state’s guarantee. This is why the sequencing argument matters more than the destination argument. Survey first, adjudication machinery second, a funded indemnity third, and the register last: a reform programme that begins with the register has begun at the wrong end.

Anirudh Burman, writing for Carnegie India in the Washington International Law Journal in 2019, proposed reaching the same destination by a different vehicle: title insurance. Drawing on the American market, where private insurers underwrite title risk after their own searches, Burman argued that insurers would do commercially what the Indian state has failed to do administratively, building clean title repositories through underwriting diligence, pricing the risk that the register cannot remove, and compensating the injured buyer from reserves regulated as insurance; he noted that the Real Estate (Regulation and Development) Act 2016 had already empowered state governments to require title insurance for real estate projects. The proposal has the merit of candour about state capacity: it does not require the universal first adjudication, it prices uncertainty instead of pretending to abolish it, and it grows from the transactions where values justify premiums. Its limit is the mirror of its merit: insurance covers the market that can pay for it, and the undocumented holder outside the formal market gains nothing. Between the Torrens destination and the insurance alternative sits the position this article’s evidence supports: the goal of a state answer to the ownership question is right; the register is the last component of that answer to build, not the first; and in the interval, the honest work is the unglamorous trio of survey, adjudication and record integration, together with duty rates low enough that transactions stop hiding from the record altogether.

The unsettled survey: the prerequisite that keeps slipping

The first prerequisite named in the obstacles analysis, a settled survey, deserves its own examination, because it is the one on which the state has been working longest and the one whose incompleteness is easiest to underestimate. A cadastral survey fixes the parcel: its boundaries, its shape, its extent, its number in the village map. India’s cadastral base was built by the settlement operations of the nineteenth and early twentieth centuries, conducted province by province with the instruments of the time, and in many districts the operative map is still, in substance, the map those operations produced, annotated by hand as holdings divided. A settlement was also a fiscal event, the occasion on which revenue demand was assessed, and the cycle of periodic resettlement that the colonial system promised largely lapsed after independence, because a fresh survey is administratively expensive and politically delicate: it reopens every boundary, corrects every extent, and can disturb the revenue demand and the possession pattern of every holder in the village at once. The result is a spatial record whose errors are not random. Subdivisions effected by sale, gift and inheritance accumulate unrecorded, because subdivision survey follows an application and a fee rather than the transaction itself; encroachments harden into boundaries by acquiescence; and the map and the ground diverge slowly, in one direction, for decades.

Modern survey method is not the constraint. The SVAMITVA scheme, launched by the Ministry of Panchayati Raj in 2020, surveys the inhabited, abadi land of villages with drones and issues property cards to the households found in occupation, an exercise that has demonstrated, at national scale, that the physical capture of parcel geometry is fast and comparatively cheap with contemporary instruments. Survey of India grade equipment, satellite imagery, and the geo-referencing work under the land records programme all point the same way: the geometry can be had. What a resurvey cannot settle by measurement is the dispute the measurement reveals. The moment a fresh survey fixes a boundary two feet inside a neighbour’s cultivation, or records an extent smaller than the revenue record has carried for forty years, the state has manufactured an adjudication, and adjudications require forums, evidence, hearings and appeals, which is the second prerequisite arriving through the door of the first. States have repeatedly chosen to digitise the old map rather than resurvey the land, and the choice is rational within their incentives: digitisation produces programme statistics and no litigants, resurvey produces accurate maps and a generation of boundary disputes. For the titling question, though, the choice is decisive, because a conclusive folio must describe a parcel, and a state that cannot say where the parcel ends cannot guarantee who owns it, whatever its register says. The survey is not a preliminary to the reform. On the Indian evidence, the survey is the reform, and the register is its epilogue.

How a deeds system actually becomes a title system: the comparative evidence

The comparative record contains one lesson that the Indian debate chronically skips, which is that no jurisdiction built a conclusive register by proclamation. England and Wales, the jurisdiction whose conveyancing vocabulary India inherited, began with voluntary registration under nineteenth-century statutes that almost nobody used, moved to compulsory registration in London at the end of that century, and extended compulsion area by area under the Land Registration Act 1925, so that compulsory first registration covered the whole of England and Wales only in 1990, roughly a century after the project began. The mechanism of conversion matters more than the dates. Registration became compulsory on the occasion of a dealing: when unregistered land was sold, the buyer had to register the title, the registry examined the deeds once, and the parcel entered the register with a class of title that could be upgraded over time as possession and the absence of claims hardened it. The register therefore grew at the speed of the market, parcel by parcel, with the examination cost borne transaction by transaction, and the state indemnity, paid out of the registry’s fee income when the register was rectified against a proprietor, accumulated its credibility over decades of small claims rather than a single founding promise. The system that resulted is conclusive with defined exceptions, rectifiable for mistake, and guaranteed by an indemnity that claimants actually recover, which is to say it is conclusive in the only sense that has ever existed anywhere: the register binds, errors are corrected prospectively, and the person injured by correction is paid.

The United States took the other fork and is the standing proof that a deeds system can persist indefinitely alongside a sophisticated land market, provided a private industry prices the risk. American recording is a deeds system of the Indian kind, county by county, giving notice and priority without any state guarantee of ownership. The gap is filled commercially by title insurance, the product Burman’s Carnegie India work proposes adapting for India, under which the insurer searches the record, excepts the defects it will not cover, and indemnifies the buyer or lender for the defects it missed. The American equilibrium has real costs, duplicated search expenditure at every transaction and premiums that function as a private tax on conveyancing, and real virtues, no founding adjudication, no public liability, and risk priced parcel by parcel. What the two comparisons establish together is that the choice is not between certainty and uncertainty. It is between three locations for the cost of uncertainty: on the state, accumulated slowly through a register and an indemnity, as in England; on the parties, priced by insurers, as in America; or on the parties, unpriced and uninsured, surfacing as litigation and discounted land values, which is the Indian arrangement. India is not outside the comparative set. It is the third cell of it.

For the Indian sequencing debate, the English path is the more instructive, because England started where India stands, with deeds, defective roots of title, and a conveyancing profession invested in search, and because the English conversion technique, compulsion triggered by dealing, examination once at first registration, graded classes of title, indemnity from fee income, answers the Indian obstacles in order. It does not demand a universal adjudication before the register opens; it lets the register open empty and fill as land transacts, with the oldest and quietest titles entering first and the disputed parcels remaining in the deeds system until their disputes resolve. It does not demand a survey of the whole country in advance; it demands a title plan for each parcel at entry. And it funds the guarantee from the transaction stream that the guarantee serves. The Rajasthan certification scheme, voluntary and urban, is a distant relative of this technique rather than of the founding-proclamation model the NITI Aayog draft follows, and the choice between those two conversion techniques, gradualist and transactional against universal and administrative, is the real policy question the titling debate should be arguing about.

The Rajasthan experiment: certification inside the old system

The first state to legislate in this direction was Rajasthan, and the character of its statute is widely misdescribed, so its design needs stating with care. The Rajasthan Urban Land (Certification of Titles) Act 2016, Rajasthan Act 12 of 2016, does not create a Torrens register and does not displace the Registration Act or the revenue record. It creates a certification service on top of them. In urban areas that the state government notifies, a holder of urban land may apply to a Certification Authority, which scrutinises the application and the documents, verifies them against the records maintained by the state and other authorities, and, if satisfied of the veracity of the information and the authenticity of the documents, issues a provisional certificate of title, recording the data in an electronic system the Act names CLEAR, the Computerized Land Evaluation and Administration of Records. The provisional certificate carries no state guarantee. It remains in force for two years, during which the holder must keep the possession and title reflected on the land, and the Authority must recall and cancel it if a bona fide counter claim or objection is filed, or if the title turns out to be in litigation before a court; no provisional certificate may issue at all where the title is disputed, in court or otherwise, and the applicant must pay certification-cum-guarantee charges and produce the prescribed no-dues certificates before certification. If the two years run without a counter claim or objection, the Authority may issue a permanent certificate of title, and to that certificate the state government stands as guarantor for the genuineness and authenticity of the title to the extent the certificate specifies, with the Act’s scheme providing for the indemnification of a person who suffers loss through a defect the certificate did not record.

Measured as implementation, the Rajasthan design is intelligent about the obstacles and constrained by them. It is voluntary, so it opens folios only where a holder wants certification and can document the claim, which is another way of saying it begins with the easiest titles. It is confined to notified urban areas, where values support the charges and the records are least decayed. It builds the objection window into the certificate itself, two years of provisional status functioning as a public invitation to rival claimants, and it excludes disputed titles at the door rather than adjudicating them. And it attaches the state guarantee only at the end of that process, to a title that has survived scrutiny and a standing invitation to object. What it cannot do follows from the same features: it does not settle the unsurveyed boundary, it does not adjudicate the disputed title, it does not reach the holder who cannot assemble documents, and its guarantee is only as funded as the charges and the state’s willingness to honour claims make it. The voluntary door also builds a selection into the scheme that any evaluation must allow for: the titles presented for certification will be disproportionately the titles already strong, held by owners with complete papers, in notified areas, against whom no rival is waiting; a certification register stocked with such titles proves that the state can guarantee easy cases, which was never the question. The hard cases, the disputed, the undocumented, the unsurveyed, remain where they were, in the presumptive system, and the two systems then coexist with the boundary between them drawn by the quality of the applicant’s documents. That is not an argument against the experiment. It is an argument for reading its results, whenever they are reported, with the selection effect in view, and against treating a successful certification scheme as evidence that the universal register is within reach. The certificate is best understood not as conclusive titling achieved but as conclusive titling rehearsed: a state learning, on a small and favourable stage, how to stand behind a title, while the registration and revenue systems continue underneath unchanged. For the series’ purposes the Act’s significance is that it exists at all. A state legislature has enacted the guarantee, defined its limits, and accepted the liability, which converts the conclusive titling debate in India from a purely comparative argument into a question about a working, if narrow, domestic precedent.

The model law route, and the replacement that changes the medium instead

The Union government’s route to the same destination has been the model law, and its two attempts bookend the Rajasthan experiment. In 2011 the Department of Land Resources released the draft Land Titling Bill, a revision of a draft first released in 2010, as a model for adoption by states, land being a state subject. The draft proposed a state Land Titling Authority, Title Registration Officers with the powers of a civil court, a Register of Titles built from survey data, a unique identification number for each property, and a record of title treated as conclusive of ownership, with objections invited and disputes routed to tribunals. No state enacted it. In 2020 NITI Aayog prepared a fresh draft, a Model Act and Rules for conclusive land titling, and circulated it to states and Union territories in June of that year for their comments, releasing it publicly in October 2020. Its architecture refines the 2011 design: the state government may order the establishment of a system of title registration for immovable properties; a Record of Titles is prepared and notified; a person aggrieved by an entry may object before the Title Registration Officer within three years of the notification; the objection is entered in a Register of Disputes and referred to a Land Dispute Resolution Officer; an appeal lies to a Land Titling Appellate Tribunal within thirty days, and onward to a special bench of the High Court; and after three years from notification the register of title attains conclusivity of its own force. The dispute institutions are deliberately temporary, designed as one-shot bodies whose work declines as the register settles, and the state guarantee carries the obligation to compensate where the conclusive register is wrong.

The model remains a model. Land and land records sit in the State List, the Centre cannot legislate a title register for the states, and adoption has to be earned state by state against the same obstacles: a state enacting the model Act inherits the survey, adjudication and indemnity problems in their full size, with the register arriving first in the statutory scheme however the implementation is sequenced. The pattern across 2011 and 2020 is worth naming, because it is the standard pattern of Indian land reform: the drafting is competent, the destination is agreed in principle by nearly every committee and commission that has examined the subject, and the binding constraint is neither drafting nor principle but the administrative and fiscal prerequisites that no model law can supply to the state that adopts it.

The Union programme’s own trajectory tells the same story from inside the administration. The 2008 programme that founded the modernisation effort named the destination in terms: its documents described a progression from computerised presumptive records to conclusive titling as the reform’s end point. The revamped programme that replaced it in 2016 is organised, in its operational components and in the statistics its ministers report to Parliament, around computerisation of the record of rights, digitisation of maps, computerisation of registration, integration of the two record systems, and the parcel identifier: the vocabulary of the destination dropped out of the targets while the targets themselves were substantially met. An implementation reading should not treat that as hypocrisy. It is the administrative system discovering, component by component, where its capacity binds, and quietly redefining success as the modernisation of the system it has rather than the construction of the system it announced. The consequence for the titling debate is that the prerequisites are being built, survey by survey and integration by integration, by a programme that no longer says it is building them, which leaves the political system without an occasion to decide the question the prerequisites lead to. The register will not arrive by drift. Some legislature, at the Centre by model or in a state by enactment, has to vote for the guarantee and fund it, and the whole record from 2011 to the NITI Aayog draft is a record of that vote being prepared for and postponed.

The proposed replacement of the 1908 Act itself belongs to a different family of reform, and confusing the two families is the commonest error in commentary on the subject. The Department of Land Resources has circulated a draft Registration Bill, in 2025, for public consultation, proposing to replace the Registration Act 1908 with a modernised statute: electronic presentation and admission of documents, online registration, electronic certificates, digital maintenance of records, an expanded list of documents requiring compulsory registration that reaches agreements to sell, powers of attorney, sale certificates and mortgages by deposit of title deeds, and identity verification that permits Aadhaar-based authentication while expressly preserving alternatives for those who do not use it. The draft is a consultation document, not an enacted law, and its status should be stated in exactly those terms. Its significance for this article is what it does not attempt. It modernises the deeds machine, bringing the statute into line with the computerised practice the states built under the old Act’s permissive provisions, and it widens the compulsory net so that fewer transactions structure themselves outside it. It does not create a register of titles, does not empower the registering officer to enquire into ownership, and does not attach a state guarantee to anything. If enacted in that form, India would have a twenty-first century deeds registration statute and the same presumptive title system it has had since the nineteenth century, which is not a criticism of the draft. It is a clarification of what the draft is for, and a warning against the headline version of it.

What a buyer should actually check

The practical payload of everything above is a discipline, not a document. Because the state will not answer the ownership question, the buyer must assemble the answer from the three records and the world outside them, and the assembly has a recognisable shape. The search begins in the registration record: an inspection of Book 1 and its indexes under Section 57, or its computerised equivalent, together with the encumbrance certificate that registration offices issue for a stated period, which lists the registered transactions affecting the property across that period. Conveyancing practice commonly runs the search back through thirty years of transfers, a convention rather than a statutory rule, adopted because a chain that holds together across a generation of deeds, successions and mortgages is materially harder to falsify than a short one, and because defects in old links tend to have declared themselves. Each link is read for what Section 17 and Section 49 teach: whether the deed was registered, whether it was executed by the person the previous link made owner, whether the description and extent stayed constant, and whether the consideration pattern suggests the duty-driven undervaluation that marks a transaction with something to hide.

Two limits of the search deserve naming, because buyers habitually over-read both documents it produces. The encumbrance certificate is a certificate about the register, not about the land: it reports the transactions the search found in the indexes for the stated period, and it is silent about everything the register never captured, including the equitable mortgage created by deposit of deeds in the seller’s bank locker and the succession that was never presented. Its value also depends on the integrity of the indexes it searches, and the indexes are organised by the names of parties, so a deed executed under a variant spelling of a name, a reality in records transliterated across scripts and generations, can escape a search conducted under the spelling the buyer was given. The root document, called the mother deed in the conveyancing practice of the southern states, carries the opposite temptation: because the chain hangs from it, a clean and registered mother deed feels like a foundation, when it is only the oldest document the parties chose to produce, and the transfers before it, by which the executant of the mother deed acquired the land, may be exactly where the defect lives. Diligence ends at a judgment about probabilities, and the professional title report, honestly written, says so in its qualifications.

The revenue record is then read against the chain, not as proof of title but as a check on it. Mutation should have followed each transfer; a gap between the last registered sale and the recorded holder is a question the seller must answer, and the answer frequently reveals the unregistered partition, the unprobated will, or the family arrangement that the deed chain conceals. The recorded extent and the survey number are checked against the deed description and against the cadastral map, and boundary or extent disagreements are resolved on the ground, by measurement, before money moves rather than after. The buyer also checks the layers no register holds: the litigation search in the courts with jurisdiction, the municipal assessment and dues position in urban areas, the approvals and tenure conditions attaching to the land, and the historical encumbrances that repealed statutes leave behind them, since urban plots still surface with declarations and proceedings from the era of the urban land ceiling legislation whose repeal closed the statute without unwinding every file it had opened. Where the search report comes back from the buyer’s counsel as a presentation setting out the chain link by link, reading it closely matters more than commissioning it, and a buyer can open and read the title search presentation with the ReportMedic presentation viewer before the advice meeting rather than during it. The general position described in this article is legal information assembled for a class of transactions; it does not substitute for advice on an individual purchase, and limitation periods and forum rules in property disputes are unforgiving of delay, so a specific doubt belongs with a practitioner instructed on the specific plot.

What this discipline cannot do should be said as plainly as what it can. A perfect search establishes that no registered claim, no recorded mutation, and no pending suit contradicts the seller’s title. It cannot establish that no claim exists. The heir who surfaces with an unregistered will, the mortgage by deposit of title deeds created by handing over documents in a notified town, the adverse possessor whose twelve years ran quietly, the prior buyer holding an unregistered agreement and possession: the search prices these risks, and the price appears in the market as the discount on rural and peri-urban land, the premium on institutional developments with insured or certified titles, and the fees of the diligence profession. That is the presumptive system working exactly as designed. The buyer is the insurer of the state’s silence, and pays the premium in diligence, in delay, or in loss.

Verdict

The Registration Act 1908 is a statute that does its job. Judged as a deeds registry, it is among the more successful pieces of the colonial legal inheritance: its compulsory net is wide, its priority rule is clear, its record is public, its authentication is fortified against personation, and its computerisation under the modernisation programme has carried the register book into the database age without breaking the legal continuity of the record. The implementation failure that surrounds Indian land title is not located in this Act’s administration. It is located in the century-long substitution of this Act’s product for a different product it never promised: a state answer to the question of ownership. The machine records deeds. India needed, and still needs, the further institutions that a guarantee of title presupposes: a settled survey, an adjudication of the claims the record cannot see, and a funded indemnity for the claims the guarantee extinguishes. Rajasthan has rehearsed the guarantee on notified urban land, NITI Aayog has drafted its architecture for the states, and the draft replacement of the 1908 Act modernises the machine without touching the question. Until the prerequisites are built, the honest description of an Indian property transaction remains the one this article began with: the state proves the document, the buyer proves the title, and the courts collect the difference.

Frequently Asked Questions

Q: What does the Registration Act 1908 require to be registered?

Section 17 of the Act makes registration compulsory for instruments of gift of immovable property, for other non-testamentary instruments that create, declare, assign, limit or extinguish a right, title or interest in immovable property of the value of one hundred rupees and upwards, for instruments acknowledging consideration for such transactions, for leases of immovable property from year to year or for a term exceeding one year or reserving a yearly rent, and for instruments transferring decrees or awards that affect such interests. Section 17(1A) adds contracts to transfer property of the kind protected by the part performance doctrine. Registration is effected by presenting the document to the Sub-Registrar of the sub-district where the property lies, ordinarily within four months of execution.

Q: Does registration in India guarantee title?

No. Registration under the Registration Act 1908 guarantees the authenticity of the record, not the truth of the ownership the document asserts. The registering officer verifies the identity of the executant and the fact of execution under Sections 32A and 34, and expressly does not enquire into the executant’s title. A registered deed therefore proves that a named person executed a document transferring the property, and it gives that document priority over unregistered documents under Section 50, but if the executant had no title, the deed conveys none. Indian title is presumptive: ownership is inferred from a chain of registered documents, revenue entries and possession, and any link in that chain can be challenged in court. No state official certifies the result.

Q: What happens if a property document is not registered under the Registration Act?

Section 49 of the Act states the consequence. A document that Section 17 requires to be registered, and that is not registered, does not affect the immovable property comprised in it and cannot be received as evidence of any transaction affecting that property. The transfer the document was meant to effect does not take place in law. The proviso preserves narrow uses: the document may still be received as evidence of a contract in a suit for specific performance, and as evidence of a collateral transaction that did not itself require a registered instrument. Since the 2001 amendment inserted Section 17(1A), an unregistered contract of the kind the part performance doctrine protects also has no effect for that purpose. In practice, an unregistered sale deed leaves the buyer with a personal claim against the seller and no title to the land.

Q: Are power of attorney property sales valid?

Not as transfers. In Suraj Lamp and Industries Private Limited v State of Haryana, (2012) 1 SCC 656, decided on 11 October 2011, the Supreme Court held that transactions structured as a sale agreement with a general power of attorney and a will do not convey title and are not a recognised mode of transfer of immovable property. A power of attorney creates an agency under the Indian Contract Act 1872; the attorney acts for the owner and acquires no ownership, however the document is worded. An agreement to sell creates a contract, enforceable through specific performance, but no interest in the land. A will operates only on death. The Court directed that such transactions not be treated as completed transfers or made the basis of mutation. Only a duly stamped and registered deed of conveyance transfers ownership.

Q: What is the difference between presumptive and conclusive title?

Under a presumptive system, which India follows, the state’s records are evidence from which ownership may be inferred, and the inference can always be displaced in court. A registered deed proves a transaction occurred; a revenue entry raises a presumption that the recorded holder holds the land; neither is binding on a claimant with a better right. Under a conclusive system, of the Torrens kind, the state maintains a register of owners for each parcel and guarantees the register. The registered owner is owner against the world, a buyer may rely on the register alone without tracing the history, and a person injured by an error in the register claims compensation from a state assurance fund instead of recovering the land. The difference is where the risk of a defective title falls: on the buyer in a presumptive system, on the state in a conclusive one.

Q: Why is so much Indian litigation about land titles?

Because the ownership question has no administrative answer and must be settled by courts. Title rests on a chain of documents that the state records without guaranteeing, on revenue entries that record possession for fiscal purposes and lag successions by years, and on survey maps that may descend from settlements generations old, and the three records routinely disagree about the same parcel. Claims that no register captures, unregistered partitions, unprobated wills, disputed successions, surface only when land is sold, mortgaged, or acquired, which converts them into litigation. The figure most often quoted, that about two-thirds of civil cases concern land, comes from the DAKSH Access to Justice Survey, which interviewed more than nine thousand litigants in three hundred district courts between November 2015 and February 2016 and found 66.2 per cent of civil cases in its sample were land or property disputes. Other counting methods give lower shares, and the estimates vary by source and method, but every method ranks land first among the families of civil disputes.

Q: What is being done to move India to conclusive land titling?

The work has proceeded on three tracks. The Digital India Land Records Modernization Programme, administered by the Department of Land Resources, has computerised records of rights, digitised cadastral maps and computerised registration offices across the states, and is integrating the registration and revenue records, which modernises the presumptive system without changing its legal character. Rajasthan enacted the first state titling statute, the Rajasthan Urban Land (Certification of Titles) Act 2016, under which titles in notified urban areas can be certified and, after a two-year provisional period without objection, guaranteed by the state. At the Union level, the Department of Land Resources circulated a draft Land Titling Bill in 2011 and NITI Aayog circulated a draft Model Act and Rules for conclusive land titling to the states in 2020, built around a register of titles that attains conclusivity three years after notification, with a state obligation to compensate. No general conclusive register operates in India.

Q: How does stamp duty affect land records?

Stamp duty on conveyances is levied ad valorem, as a percentage of the consideration or market value, at rates fixed by states, and it is heavy enough to shape the behaviour the record then reflects. Because duty is charged on the figure the deed states, parties understate the consideration, the balance moves outside the document, and the registered record of prices becomes unreliable; states answer with circle rates and guideline values that fix a floor for assessment. Because duty is charged on the conveyance itself, parties avoid the conveyance, structuring sales as agreements with powers of attorney and wills, which keeps the transaction off the register entirely. The Supreme Court in Suraj Lamp identified high duty as the engine of both practices and urged lower rates. The quality of the land record is therefore partly a fiscal outcome: a duty structure that rewards concealment produces records that conceal.

Q: What does a Sub-Registrar check before registering a sale deed?

The Sub-Registrar’s enquiry is prescribed by the Act and is deliberately confined. Under Sections 32 and 32A, the officer checks that the document is presented by a person entitled to present it, and takes the passport-size photographs and fingerprints of the presenter and, for a transfer of ownership, of each buyer and seller named. Under Section 34, the officer enquires whether the document was executed by the person it purports to be executed by, satisfies himself as to the identity of those appearing and alleging execution, and checks the authority of anyone appearing as representative, assign or agent. The officer also confirms presentation within the time Section 23 allows and payment of the proper stamp duty and fee. What the officer does not check is the seller’s ownership, the existence of encumbrances, or the truth of the price: the enquiry certifies execution, not title.

Q: What is mutation of land records, and does mutation prove ownership?

Mutation is the updating of the revenue record, the record of rights, to reflect a change in the holding of land: a sale, a gift, an inheritance, or a partition. It is an administrative and fiscal act performed so the state knows from whom to collect revenue and to whom to send its notices, and in many states a registered transfer triggers the mutation process automatically where the registration and revenue systems are integrated. Mutation does not prove ownership. In Sawarni v Inder Kaur, (1996) 6 SCC 223, the Supreme Court held that mutation entries neither create nor extinguish title and carry no presumptive value on the question of ownership; they serve the fiscal purpose for which they are made. The recorded holder may hold for a family, as a trustee, or wrongly. Mutation is a check a buyer performs on the seller’s story, not a substitute for the title search.

Q: How far back should a title search go when buying property in India?

There is no statutory period. The governing consideration is that the search must reconstruct an unbroken chain from a root of title the buyer can trust to the seller, testing every link for registration, identity of the executant with the person the previous link made owner, and consistency of description and extent. In conveyancing practice, searches commonly run back through thirty years of transfers, and lenders and their counsel adopt similar periods when issuing title reports. The convention reflects two judgments: that a chain holding together across a generation of deeds is materially harder to falsify, and that defects tend to declare themselves over time through claims, litigation, or inconsistent entries. Older roots matter where the chain contains them: a government grant, a partition, or a succession within the period must be examined on its own documents, not passed over because the deeds on either side of it are registered.

Q: Can an unregistered agreement to sell be used in a suit for specific performance?

Yes, within the limit Section 49 itself sets. Section 49 bars an unregistered document that was required to be registered from affecting the property or being received as evidence of a transaction affecting it, but the proviso expressly allows such a document to be received as evidence of a contract in a suit for specific performance under the Specific Relief Act. The buyer under an unregistered agreement may therefore sue to compel the seller to execute and register a proper conveyance, producing the agreement as proof of the contract. What the buyer cannot do is treat the agreement as itself a transfer, claim ownership under it, or resist the seller’s transferee on its strength. The proviso also permits use of the document as evidence of a collateral transaction not required to be effected by a registered instrument, such as the fact or character of possession.

Q: How much time does the Registration Act allow for presenting a document for registration?

Section 23 sets the ordinary window: a document other than a will must be presented for registration within four months of its execution. Section 24 adjusts the rule where several persons execute the same document at different times, running the four months from each execution for the purposes of presentation and re-registration. Where delay results from urgent necessity or unavoidable accident, Section 25 empowers the Registrar to direct that the document be accepted although the further delay does not exceed four months, on payment of a fine not exceeding ten times the amount of the proper registration fee. A document executed outside India may be accepted if presented within four months of its arrival in India, under Section 26. Wills stand outside the discipline: Section 27 permits a will to be presented or deposited at any time. Presentation after the permissible period is refused, and the unregistered document then falls under Section 49.

Q: What is the Torrens system of title registration?

The Torrens system is the model of title registration devised by Sir Robert Richard Torrens and enacted in South Australia by the Real Property Act 1858, later adopted across Australia, New Zealand and parts of the Commonwealth. Instead of registering documents, the state registers titles: each parcel has a folio in the register naming its owner and its encumbrances, the state examines the claim before the first entry, and thereafter the folio is conclusive. A transfer is made by entry on the folio, not by a deed whose validity the buyer must investigate, and the history behind the register is kept behind a curtain the buyer need not lift. Because conclusiveness can injure a person whose right the register omits, the system carries an assurance fund, financed from registration fees, from which such a person is compensated. The system’s three principles are called the mirror, the curtain and the insurance: the register reflects the title completely, the past is hidden behind it, and the fund answers for its errors.

Q: What does the Rajasthan Urban Land (Certification of Titles) Act 2016 do?

The Act, Rajasthan Act 12 of 2016, sets up a title certification service for urban land in areas the state government notifies, the first statute of its kind enacted by an Indian state. A holder applies to a Certification Authority, which scrutinises the documents, verifies them against government records, and issues a provisional certificate of title, recorded in an electronic system the Act names CLEAR. No certificate issues where the title is disputed, and a provisional certificate is recalled and cancelled if a bona fide counter claim or objection arrives or the title is found to be in litigation. The provisional certificate runs for two years without any state guarantee. If that period passes with no counter claim or objection, the Authority may issue a permanent certificate, for which the state government stands as guarantor of the genuineness and authenticity of the title to the extent the certificate specifies, and the scheme provides for indemnifying a person who suffers loss through a defect the certificate did not record.

Q: What is the Unique Land Parcel Identification Number or Bhu-Aadhaar?

The Unique Land Parcel Identification Number, called ULPIN and described by the Department of Land Resources as Bhu-Aadhaar or an Aadhaar for land, is a fourteen-digit alphanumeric identifier assigned to a land parcel and generated from the geo-coordinates of the parcel’s vertices. It is a component of the Digital India Land Records Modernization Programme, and its purpose is to give every parcel a single stable identity that the registration record, the revenue record and the survey map can share, so that the three records can be linked and a change in one system can be followed in the others. The Minister of State for Rural Development told the Lok Sabha, on figures from the programme’s management information system as on 9 February 2026, that more than thirty-eight crore land parcels had been assigned the number. The identifier solves the problem of parcels being described differently across records; it does not certify the ownership of the parcel it identifies.

Q: Why do the registration record, the revenue record and the survey map disagree about the same plot?

Because they were built to answer different questions, by different departments, and they are updated by different events. The registration record, kept under the Registration Act 1908, records documents: it changes when a deed is presented, and it says nothing about possession or extent. The revenue record records the holder for revenue purposes: it changes through mutation, which follows sales and successions only when the change is reported and processed, and entries can lag the events by years. The survey record records geometry: it changes at settlement and resurvey operations, which in many districts are generations apart, so partitions and subdivisions recognised by everyone on the ground may never reach the map. Add three further sources of divergence, the understatement of consideration that stamp duty encourages, the unregistered family arrangements that villages act on without paper, and the drift of names and survey numbers across a century of copying, and disagreement between the records is the normal condition. A buyer who finds the three records in agreement has found something worth remarking on.

Q: How can a buyer inspect past registrations of a property before buying it?

Section 57 of the Registration Act 1908 makes the register a public record in a practical sense. It obliges the registering officer to allow any person applying to inspect Book 1, the register book in which documents relating to immovable property are copied, and the indexes relating to it, and to give certified copies of entries to applicants on payment of the prescribed fee. Copies given under Section 57 carry statutory value for proving the contents of the original documents. In practice the buyer, or the buyer’s counsel, applies for inspection and for an encumbrance certificate for a chosen period, which the registration office issues listing the registered transactions affecting the property in that period; in states that have computerised registration, the search runs through the office database and certified copies issue from the digital record. The inspection reconstructs the registered chain. It cannot disclose transactions that were never presented, which is why the search is combined with the revenue record, the survey map and a litigation search.

Q: Is it compulsory to register a will under the Registration Act 1908?

No. Section 17, which imposes compulsory registration, reaches non-testamentary instruments, and a will is a testamentary instrument that takes effect only on the testator’s death. Section 18 lists wills among the documents whose registration is optional, and registration gives the will the ordinary benefits of the register: a preserved copy in the custody of the registration office, protection against loss, suppression or tampering with the original, and certified copies under Section 57 for those entitled to them. Section 27 permits a will to be presented for registration, or deposited with the Registrar under the deposit provisions of the Act, at any time, free of the four-month discipline that governs other documents. The optionality of the will is one of the structural gaps in the deeds system: the single most common instrument by which Indian property changes hands at death is the instrument the compulsory record captures least.

Q: What would a state need in place before it could guarantee conclusive land titles?

Three prerequisites, none of which a titling statute can create by its own force. The first is a settled survey: parcels identified, boundaries fixed on the ground and the map, subdivisions recorded, so the state knows what physical thing each title describes. The second is an adjudication mechanism capable of resolving the competing claims that a first registration would otherwise extinguish: the unmutated successions, unregistered partitions, possessions and customary interests that no deed records, heard and decided once, fairly and at scale, because conclusiveness converts the losers’ claims against the land into claims against the state. The third is a funded indemnity: a guarantee is a promise to pay when the register is wrong, and it is credible only if an assurance fund holds money measured against the land values guaranteed. A state that builds the register before the survey, the adjudication and the fund has not abolished title uncertainty; it has given the errors the force of law and the state’s name.