Every contested trial in India is, before it is anything else, an argument about what the judge is allowed to hear. The parties arrive with a story, with documents, with recordings, with witnesses who remember and witnesses who have been rehearsed, and between all of that material and the court’s mind stands a filter built in 1872. The Indian Evidence Act is that filter. It does not define offences, create rights, or prescribe punishments. It performs the prior and quieter function of deciding what counts as material from which a court may reason at all, and on what terms the material that passes must then be proved. A litigant can hold the decisive fact in hand, in the form of a confession, a letter, a recording, or a dying man’s last words, and still lose everything at the filter, because the confession was made to a police officer, the letter is a copy whose original nobody can account for, the recording carries no certificate, or the words were spoken by someone whose knowledge the Act does not allow a court to use. The statute is best understood not as a list of rules but as a sequence of gates, and this article works through those gates in the order in which evidence meets them.

The sequence matters because each gate asks a different question, and collapsing the questions is the single most common error in arguments about this statute. The first question is relevancy: is this fact the kind of fact the law permits a court to reason from at all? The second is admissibility in the strict sense: even among relevant facts, does some specific rule of exclusion, privilege, or form keep this item out? The third is proof: the item having been let in, has it actually been established in the manner the Act requires, by primary evidence where primary evidence is demanded, by a witness who speaks from personal knowledge, under a burden of proof that someone had to discharge? A fact can be relevant and yet inadmissible, as a confession to a police officer is. A document can be admissible in form and yet unproved, as a photocopy casually tendered without the groundwork the Act requires is. A fact can sail through both gates and still fail at appreciation, the final weighing in which a court decides what the admitted and proved material is worth. The Act was drafted to keep these operations separate, and its three parts map onto them with unusual fidelity. That architecture, and the case law that has grown over its pressure points, is the subject of what follows. The companion codes that supply the substantive and procedural halves of a criminal trial, the Indian Penal Code on the side of liability and the criminal procedure code on the side of process, both presuppose this filter at every step, because neither of them decides what a court may hear.

What Stood Before the Act, and Why It Failed

Before 1872, the law of evidence in British India was not one law at all. In the presidency towns, courts applied English evidence law in a recognisable form. In the mofussil, the courts of the interior, the position was a patchwork: English rules applied so far as they suited local conditions, regulations made piecemeal by the presidencies, and, for parties professing them, survivals of Hindu and Muhammadan law on proof and oath. The result was that the admissibility of the same item of evidence could turn on where the suit was filed and who the parties were. A system that aspired to uniform justice administered through a single hierarchy of courts could not indefinitely run a rulebook that changed with the district. Codification was the characteristic nineteenth-century answer to that problem across Indian law, and evidence was always going to be one of its subjects, because evidence is the part of the law that every other part depends upon.

The drafting history explains the shape the codification took. An earlier attempt associated with Henry Maine, who served as Law Member of the Viceroy’s Council through most of the 1860s, produced a draft that was set aside as unsuited to Indian conditions, an exercise in stating English law rather than in designing a code for the courts that would actually use it. James Fitzjames Stephen, who succeeded Maine as Law Member in 1869, approached the task differently. Stephen distrusted the sprawling case-law form of English evidence law, with its exceptions layered over exceptions, and believed the subject could be reduced to a compact set of propositions organised around a single organising idea, the relevancy of facts. He prepared the draft that, introduced in 1871 and passed the following year as Act 1 of 1872, became the Indian Evidence Act, in force from 1 September 1872. Stephen later distilled the same scheme into his English Digest of the Law of Evidence, but the Indian statute came first, and it is the statute, not the digest, that has had a century and a half of consequences.

The codifying ambition carried a deliberate wager. English evidence law in 1872 was a living body of doctrine that judges developed case by case; Stephen froze a version of it into 167 sections and wagered that the frozen version would be clear enough to be applied by a judiciary of very uneven training across an enormous territory, and durable enough not to need constant legislative repair. The wager largely paid off. The Act’s structure survived with its numbering and its central concepts intact for the whole of its working life, which is a record few codes of any country can match, and the amendments that did accumulate were mostly additions at identifiable pressure points, presumptions for new social harms in the 1980s, a consent presumption in 2013, and the electronic records provisions of 2000, rather than revisions of the core design. The price of the wager is equally visible: where social or technological change outran the text, the adaptation had to be done by judges interpreting a nineteenth-century code, and the strain shows most plainly exactly where a modern reader would predict, in the law of confessions extracted by police and in the law of electronic records. Both pressure points receive extended treatment below, because both are places where the record of litigation, and not the bare text, tells the reader what the law is.

The Architecture of the Indian Evidence Act: Three Parts, Three Gates

The Act extends to the whole of India and applies to all judicial proceedings in or before any court, including courts martial, subject to exceptions for courts martial convened under the service Acts governing the army, navy and air force. Two exclusions in Section 1 do quiet but consequential work. The Act does not apply to affidavits presented to a court or officer, which is why the contents of an affidavit are governed by a different discipline from the contents of testimony, and it does not apply to proceedings before an arbitrator, which is why arbitral tribunals work with a looser evidential hand than courts while remaining free to borrow the Act’s logic. Within its field, though, the Act claims completeness of a specific kind: Section 5 provides that evidence may be given of facts in issue and of facts declared by the Act to be relevant, “and of no others”. The statute is therefore not a set of permissions layered over a general freedom to prove anything persuasive. It is a closed system in which an item enters only through a door the Act has built.

The arrangement of the 167 sections into three parts and eleven chapters is the clearest statement of the three-gate theory. Part I, Sections 1 to 55, deals with the relevancy of facts: the preliminary provisions and the great second chapter that declares which classes of fact are relevant. Part II, Sections 56 to 100, deals with proof: facts that need not be proved at all, oral evidence, documentary evidence, and the rules that exclude oral evidence where a document governs. Part III, Sections 101 to 167, deals with the production and effect of evidence: who must prove what, estoppel, witnesses, their examination, and the consequences of getting admissibility wrong. A reader who holds this map can locate any provision in the statute by asking which operation it performs, and a reader who cannot say which part a rule sits in usually does not yet understand what the rule is for.

Two definitional provisions carry more weight than their position at the front of the Act suggests. Section 3 defines evidence itself as comprising exactly two things: the statements a court permits or requires witnesses to make before it, called oral evidence, and all documents, including electronic records, produced for the inspection of the court, called documentary evidence. Anything that is neither a witness statement nor a document produced for inspection is not evidence under the Act, however persuasive it may be as a matter of common sense, and the long-running puzzles about objects, site inspections and demonstrations are resolved by forcing them through one of these two doors or treating them as aids to understanding what came through a door. Section 3 also defines a fact as proved when, after considering the matters before it, the court either believes it to exist or considers its existence so probable that a prudent person ought, in the circumstances of the case, to act on the supposition that it exists. That prudent person standard, stated in 1872, is the Act’s quiet definition of the standard of proof itself, and it is calibrated by context: the probability that suffices for a prudent person deciding a civil claim is not the probability that suffices before sending a person to prison, a calibration the courts have always read into the definition.

The three questions of the filter then present themselves in order. Is the fact relevant, meaning does it fall within one of the classes Sections 6 to 55 declare relevant, or is it a fact in issue itself? Is it admissible, meaning does it survive the exclusionary rules, the privileges, and the formal requirements for its mode of proof? And is it proved, meaning has the party who bears the burden on that fact actually established it by evidence the Act accepts, to the standard the prudent person formulation implies? The namable claim of this article can be stated plainly: relevancy is not admissibility and neither is proof, and most failed arguments about evidence in India fail because one of these three operations has been silently substituted for another. The confession that is plainly relevant and rightly excluded, the photocopy that is formally received and never proved, the dying declaration that clears every gate and is then disbelieved, each is a different failure at a different gate, and the remedies for the three failures are entirely different.

Who decides whether a piece of evidence is admissible?

Under Section 136, the judge decides. When a party proposes to give evidence of a fact, the judge may ask how the fact, if proved, would be relevant, and admits it only if satisfied on that point; where admissibility depends on another fact, the judge may require the foundation to be proved first.

Admissibility is therefore a question of law for the court, decided at the gate, and not a matter of agreement between the parties or of weight assessed later.

That allocation of the decision has practical consequences that run through every trial. Because the judge controls the gate, an objection to admissibility must be taken when the evidence is tendered, and evidence received without objection generally stays on the record to be weighed, a position the Supreme Court has repeated in the electronic records context when parties tried to raise certificate objections for the first time on appeal. Section 136 also gives the judge the tool for handling conditional relevance, the everyday situation in which a statement is relevant only if some connecting fact is true: the court may admit the statement on an undertaking that the connection will be proved, or require the connection first. The provision is short, and experienced trial lawyers treat it as the most operationally important section in Part III, because it converts the abstract three-gate theory into a daily sequence of rulings at the moment evidence is offered.

Proved, Disproved, and Not Proved: The Act’s Verdict Vocabulary

Section 3 supplies two further definitions that trial lawyers use daily and lay accounts of the Act usually skip. A fact is said to be disproved when, after considering the matters before it, the court either believes that it does not exist or considers its non-existence so probable that a prudent person ought, under the circumstances of the particular case, to act upon the supposition that it does not exist. A fact is said to be not proved when it is neither proved nor disproved. The third state is the operative one in most litigation, because most contested facts end exactly there: the court is not persuaded either way, and the case is then decided not by belief but by the burden provisions, which exist precisely to allocate the loss in the not proved state. This is why the burden chapter of Part III is not an appendix to the law of proof but its engine. A judge who cannot decide what happened does not have the option of declining to decide; the Act tells the judge which party’s failure the indecision counts against, fact by fact, and the verdict is built out of those allocations.

The calibration of the prudent person standard between civil and criminal proceedings is performed by the courts rather than the text, and two received formulations state it. In criminal cases, the prosecution’s burden carries the quality associated with the golden thread of English criminal law, the principle of Woolmington v Director of Public Prosecutions, that the prosecution must prove the prisoner’s guilt and that no burden rests on the accused to establish innocence; Indian courts received the principle as consonant with Sections 101 to 104 and quote it still. Within that burden, the doctrine of benefit of doubt does the final work: where the evidence leaves two views reasonably possible, the view favourable to the accused is to be adopted, a proposition the Supreme Court stated in Kali Ram v State of Himachal Pradesh in 1973 while warning, in the same line of authority, that the doubt must be a reasonable doubt arising from the evidence and not an imaginary or fanciful hesitation. In civil proceedings the same statutory words are worked at the preponderance setting, so that a fact is proved when the court finds its existence more probable than its non-existence on the evidence as a whole. The Act’s single definition thus supports two disciplines, and a reader who attends to the prudent person formula, rather than to slogans about standards, can see how: the prudent person acts differently when the act in contemplation is the taking of a citizen’s liberty than when it is the award of money between merchants, and the definition builds the circumstances of the particular case into the standard itself.

The Evidence Filter: Five Stages Between a Fact and a Finding

Stage	Governing provisions	The question the court asks	Where evidence commonly fails
Relevancy	Sections 5 to 55	Is this fact one the Act permits a court to reason from at all?	The fact is persuasive in ordinary life but falls outside every declared class of relevant fact.
Admissibility	Sections 24 to 27, 91 to 100, 122 to 129, 136	Does an exclusionary rule, a privilege, or a formal requirement keep this item out despite its relevance?	The confession was made to a police officer, or the oral term contradicts the written contract that governs the matter.
Mode of proof	Sections 59 to 90A, 65A and 65B	Has the item been proved in the manner the Act demands for its kind?	A copy is tendered where primary evidence is required, or an electronic record is produced without the certificate Section 65B demands.
Burden and presumption	Sections 101 to 114A	Who had to prove this fact, and does a presumption shift or discharge that burden?	The party bearing the burden led no evidence on the fact, or a mandatory presumption was ignored.
Appreciation	Sections 3, 114, 134, 165, 167	What is the admitted and proved material worth, taken as a whole?	The evidence passed every gate and is then disbelieved, or a sole witness is found unreliable on close scrutiny.

The First Gate: Relevancy Under Part I

Part I opens with the general proposition of Section 5 and then spends fifty sections declaring classes of fact to be relevant, and the classes repay grouping by the kind of connection each asserts between the fact offered and the fact in issue. The earliest sections work with connection in time, place and circumstance. Section 7 makes relevant the facts that are the occasion, cause or effect of facts in issue or relevant facts, the physical setting from which the event grew and the traces it left behind. Section 9 admits facts necessary to explain or introduce a relevant fact, facts that support or rebut an inference, facts that establish the identity of a person or thing, and facts that fix the time or place at which something happened, the quiet workhorse provisions under which identification evidence and much of the furniture of a trial enter. Section 11 performs the reverse operation: facts that are inconsistent with a fact in issue, or that make its existence or non-existence highly probable or improbable, are relevant, which is the statutory home of the plea of alibi.

What is res gestae under Section 6 of the Indian Evidence Act?

Section 6 makes relevant any fact so connected with a fact in issue as to form part of the transaction, whether it occurred at the same time and place or at different times and places. The doctrine, called res gestae, admits the spontaneous words and acts belonging to the event itself, such as the cries of bystanders during an assault.

The same-transaction test is deliberately elastic, and its elasticity is the point of dispute in case after case. A statement made during the event, or so shortly after it that the speaker is still within the event’s grip, comes in as part of the transaction; the same statement made hours later, after reflection and the opportunity to fabricate, is a narrative about the event and must find some other door or stay out. The illustrations to the section work with a beating witnessed by bystanders whose words, spoken as the blows fell, are part of the transaction. Courts applying Section 6 are therefore always asking a question about spontaneity and continuity rather than about the clock alone, and the section’s overlap with the dying declaration provisions and with Section 8 conduct is managed by remembering that one item of material can pass the relevancy gate through several doors at once.

Section 8, the provision the brief for this article singles out, makes relevant three things that surround almost every disputed act: motive, preparation, and conduct. The fact that a person had a motive for the act charged is relevant, as is the fact that the person made preparations for it, buying the poison, hiring the vehicle, reconnoitring the place. Conduct, previous or subsequent, is relevant where it influences or is influenced by a fact in issue or a relevant fact, and the section’s explanations extend the idea to statements that accompany and explain conduct, and to statements by a person against whom an offence is committed where they amount to a complaint made shortly after the offence. Subsequent conduct of the most eloquent kind, flight from the scene, concealment of the weapon, silence in the face of accusation where a denial would be natural, all enters through Section 8. Section 10 adds the conspiracy rule: where there is reasonable ground to believe that two or more persons have conspired together, anything said, done or written by any one of them in reference to their common intention, after the time the intention was first entertained, is relevant against each of them, a provision of remarkable breadth whose threshold requirement, reasonable ground to believe in the conspiracy before the provision unlocks, is the control on its use.

The middle of Part I governs statements, and statements enter under a set of distinct regimes that this article takes in turn: admissions and confessions, Sections 17 to 31, receive their own discussion below because of the police confession problem; statements by persons who cannot be called as witnesses, Sections 32 and 33, are headed by the dying declaration and also receive separate treatment. Judgments occupy Sections 40 to 44, and their logic is restrictive in a way that surprises lay readers. A previous judgment is relevant mainly where it bars a second suit or trial, under Section 40. Section 41 gives special force to final judgments of competent courts in probate, matrimonial, admiralty and insolvency matters, judgments in rem that pronounce on a legal character or status against the world. Judgments on matters of a public nature relevant to the inquiry enter under Section 42, but Section 43 then states the default with characteristic bluntness: judgments other than those mentioned in the preceding sections are irrelevant unless the existence of the judgment is itself a fact in issue or is relevant under some other provision. The prior conviction of a co-accused, the findings of a civil court in a parallel dispute, the opinion of another judge about the same event: none of these is evidence of the facts found, because the Act treats judgment as the conclusion of a process between particular parties and not as a portable fact about the world. Section 44 completes the set by allowing a party to show that a judgment relied on against it was delivered by a court without competence, or obtained by fraud or collusion.

Opinions form the last great class in Part I, Sections 45 to 51, and their premise is a division of labour: witnesses speak to facts, and opinion is in general the court’s alone, so the sections admit the opinions of third persons only in defined enclaves where the court acknowledges it needs help. Section 45, the head of the family, makes relevant the opinions of experts on points of foreign law, of science or art, and on the identity of handwriting or finger impressions, when the court must form an opinion on such a point. Around it sit the opinion of the Examiner of Electronic Evidence under Section 45A, facts that bear upon an expert’s opinion under Section 46, opinions on handwriting under Section 47 and on electronic signatures under Section 47A, opinions on the existence of rights and customs under Section 48, on usages and tenets under Section 49, on relationship under Section 50, and the grounds of any admitted opinion under Section 51. Character closes Part I, Sections 52 to 55, on an asymmetrical scheme: in civil cases, character offered to prove the probability of imputed conduct is generally irrelevant under Section 52; in criminal cases, the accused’s previous good character is relevant under Section 53 while previous bad character is, under Section 54, irrelevant except in reply where good character has been asserted or where character is itself in issue; and Section 55 admits character as affecting the amount of damages in civil claims. The asymmetry is a deliberate protection: the accused may open the door to character, and the prosecution may not force it open first.

Opinions and Identification: Two Places Where the Gate Widens

The opinion sections deserve a closer reading than their position at the end of Part I suggests, because the expert witness is the point at which the Act’s division of labour between witness and court is consciously relaxed. The ordinary witness is confined to facts perceived; the expert is invited to do the very thing forbidden to everyone else, to state an inference, because the court acknowledges that on a point of foreign law, science or art, or the identity of handwriting or finger impressions, it cannot draw the inference competently on its own. Section 45 makes the expert’s opinion a relevant fact, and Section 46 admits facts, not otherwise relevant, that support or contradict the expert’s opinion, which is where the data underlying the opinion, comparable specimens, experiments and results in similar cases, enter. The discipline surrounding the concession is what keeps it from swallowing the trial. An expert’s opinion is not proof of the facts on which it rests; those facts must themselves be proved. The opinion is advisory in character, and its value depends on the reasoning exposed to the court: the qualifications of the expert, the material examined, the process followed, and the candour with which the opinion survives cross-examination directed precisely at those foundations. Courts therefore treat a bare conclusion, a signature and a designation, as worth little, and treat an opinion that explains itself as worth weighing, a practice so consistent that it functions as doctrine: the expert assists the court, and an expert who argues a side has stepped out of the only role the section gives him.

Two recurring applications mark the boundaries. Medical opinion stands against eyewitness account in every injury trial, and the received rule, stated by the Supreme Court in State of Haryana v Bhagirath in 1999, is that medical evidence is essentially opinion evidence, valuable and often decisive on the nature of injuries and the weapon or mechanism that could have caused them, but not conclusive against a trustworthy ocular account merely because some variance appears between the two; where the medical evidence completely rules out the possibility of the injury occurring in the manner the eyewitnesses describe, the position reverses, and the ocular account cannot stand. Handwriting opinion occupies the other boundary: the court may act on the opinion of a handwriting expert, and may also compare the disputed writing with admitted specimens itself under Section 73, but the cases counsel caution in resting a finding on expert comparison alone, since handwriting expertise is among the least exact of the forensic disciplines the section admits. The electronic counterpart, the Examiner of Electronic Evidence under Section 45A, carries the same structure into the digital record, as the electronic evidence discussion below shows: the opinion is a relevant fact, the underlying facts must still be proved, and the certificate gate must be passed before the opinion question even arises.

Identification evidence enters at the other end of Part I, through Section 9, and it illustrates how the Act lets a fact in quietly while the courts build the safeguards at the appreciation stage. Section 9 makes relevant facts that establish the identity of any person or thing whose identity is in question, and the dock identification of the accused by a witness in court is evidence under the section. The investigative device that supports it, the test identification parade, in which a witness is asked to pick out the suspect from among similar persons during the investigation, has no separate statutory home in the Act; its product reaches the court through Section 9 and the testimony of the witness and of the officer or Magistrate who conducted the parade. The governing approach, articulated by the Supreme Court in Budhsen v State of Uttar Pradesh in 1970, treats the parade as part of the investigation whose purpose is to test and strengthen the trustworthiness of the witness’s identification, so that the substantive evidence remains the identification in court and the parade furnishes corroboration of it. The factors the courts weigh are practical and have hardened into a checklist through repetition: the witness’s opportunity to observe the offender, the light and distance, the interval between the event and the parade, the disclosure or non-disclosure of identifying features before the parade, and whether the accused was previously known to the witness, in which case the parade proves nothing. The pattern is characteristic of the Act’s method and worth noticing at this point in the article: the text opens a wide gate at relevancy, and the case law, rather than narrowing the text, builds a discipline of weight at the far side of it.

Admissions, Confessions, and the Closed Door to the Police Station

Admissions and confessions are the Act’s most heavily litigated statements, and the Act builds the regime in layers. An admission, defined by Section 17, is a statement, oral or documentary or in electronic form, which suggests an inference as to a fact in issue or a relevant fact, made by one of the persons Section 18 specifies, principally the parties to the proceeding and persons whose position or liability is in question. An admission is relevant and may be proved against the person who made it, but the maker may not ordinarily prove it in his own favour, Section 21 states the rule with its exceptions, since a self-serving statement manufactured after a dispute arose would otherwise flood every trial. Section 23 protects the negotiating table in civil cases: an admission made upon an express condition that it will not be used, or in circumstances from which the court infers the parties agreed it would not be used, is not relevant, which is the statutory kernel of the without prejudice rule. A confession is a species of admission, and although the Act nowhere defines the word, the Privy Council in Pakala Narayana Swami v Emperor in 1939 fixed the meaning that has governed since: a confession must either admit the offence in terms or admit substantially all the facts that constitute the offence, and a statement that contains self-exculpatory matter, which if true would negate the offence, is not a confession at all. The distinction matters because the exclusionary rules that follow are aimed at confessions, and statements falling short of confession are judged by the milder law of admissions.

The first exclusion is Section 24. A confession appears to the court to have been caused by an inducement, threat or promise, having reference to the charge, proceeding from a person in authority, and sufficient in the court’s opinion to give the accused grounds to suppose he would gain an advantage or avoid an evil of a temporal nature in reference to the proceedings, is irrelevant. The provision is aimed at reliability through a specific channel: it is the confession extracted by hope or fear held out by authority that the law distrusts, because a person bargaining for advantage may confess to what did not happen. Section 28 restores a confession after the impression of the inducement has been fully removed, and Section 29 preserves confessions otherwise relevant against objections that they were made under a promise of secrecy, in consequence of deception, while drunk, in answer to questions the accused need not have answered, or without the warning that he was not bound to confess, a list that marks the boundary of the inducement rule with some precision. Section 30 then permits the court, where more persons than one are being tried jointly for the same offence and a confession by one affecting himself and another is proved, to take the confession into consideration against the other as well as against its maker.

Can a confession by one accused be used against a co-accused?

Section 30 allows the court, in a joint trial for the same offence, to take a proved confession of one accused into consideration against a co-accused whom it also implicates. It is not evidence in the strict sense against the co-accused and cannot by itself found a conviction.

The courts have consistently required the prosecution case against the co-accused to stand on other evidence, with the confession lending assurance to conclusions that the other evidence already supports.

Sections 25 and 26 then close the door that dominates Indian criminal practice. Section 25 states the rule without qualification: no confession made to a police officer shall be proved as against a person accused of any offence. Section 26 extends the exclusion to any confession made by a person while in the custody of a police officer, to anyone at all, unless made in the immediate presence of a Magistrate. The two sections together create the distinctive Indian position, more absolute than the English rule from which the Act otherwise descends, that the police station is an evidential dead zone for confessions: a person may confess to the investigating officer in the clearest terms, and the confession simply does not exist for the trial. The route the law provides instead is the recorded confession before a Magistrate, with the safeguards of warning and reflection time that the criminal procedure code attaches to it, and a reader working through the recording machinery will find it in the pillar treatment of the Code of Criminal Procedure 1973. The rationale usually given for the bar is institutional distrust hardened into rule: confessions to police are too easy to extract by pressure and too difficult to test afterwards, since the contest at trial would collapse into oath against oath between the accused and the officer, and the legislature of 1872 chose to sacrifice the reliable confessions along with the coerced ones rather than litigate the difference in every case.

Section 27 is the proviso that reopens a narrow part of what Sections 25 and 26 close, and it is the most debated provision in the Act. Its text provides that when any fact is deposed to as discovered in consequence of information received from a person accused of an offence, in the custody of a police officer, so much of the information, whether it amounts to a confession or not, as relates distinctly to the fact thereby discovered, may be proved. The Privy Council in Pulukuri Kottaya v Emperor in 1947 gave the provision its enduring construction: the fact discovered is not the object produced but the place from which the object is produced and the accused’s knowledge of it, the information admitted must be severed so that only the part relating distinctly to the discovery comes in, and words of past use or past history, the words “with which I stabbed him” in the illustration the judgment worked with, stay out. The Supreme Court in Mohd. Inayatullah v State of Maharashtra in 1976 confirmed that severance is the court’s duty, admitting the portion of the statement that was the immediate and direct cause of the discovery and rejecting the rest. The constitutional attack on the section, that it discriminated between accused persons in police custody and others for the purposes of Article 14, was repelled by the Supreme Court in State of Uttar Pradesh v Deoman Upadhyaya in 1960, which accepted the discovery guarantee as a rational basis for the distinction.

The standing argument about Section 27 should be stated in its two strongest forms, because the stance of this series on contested provisions is to let each side speak at full strength before any verdict. For the provision, the argument is the one the Privy Council articulated: the discovery itself is the guarantee of truth of the portion admitted. A statement that leads police to a concealed weapon in a place only the person connected with the crime would know corroborates itself by the physical fact it produces, and the narrowness of the exception, confined to the words that relate distinctly to the discovered fact, keeps the confession bar intact for everything else. Against the provision, the argument is that the exception reverses the incentive structure Section 25 was built to create: if a custodial statement can still yield admissible evidence whenever it is dressed as information leading to a discovery, the pressure to extract statements in custody returns through the discovery door, and the reliability of the discovery itself depends on police testimony about where the accused led them and what he said, the very oath-against-oath contest the bar was designed to avoid. The same line of argument presses the artificiality of severing a sentence mid-course and pretending that the admissible fragment carries no confessional colour for the court that hears it. The verdict the record supports is narrower than either side’s rhetoric. Section 27 has survived a constitutional challenge and a century and a half of use because its core case, the genuinely new fact discovered at the accused’s instance, is probative in a way no other evidence can replicate; but its integrity depends entirely on courts insisting, as Kottaya and Inayatullah insisted, on genuine discovery, on severance, and on rejecting information about facts the police already knew, and the reported cases in which discoveries were staged or pre-known show what the provision becomes when that insistence slackens.

Judicial and Extra-Judicial Confessions: The Two Surviving Channels

The confession bar leaves two channels through which a confession can still reach the court, and each has its own pathology. A judicial confession is one recorded by a Magistrate in the course of an inquiry or trial, under the recording provision of the criminal procedure code, with the warnings the code prescribes: the Magistrate must explain to the person that he is not bound to confess and that the confession, if made, may be used as evidence against him, and must satisfy himself that the confession is voluntary, allowing time for reflection and keeping the person out of police influence in the interval. A confession recorded with those safeguards is strong evidence, and the contests around it usually concern whether the safeguards were genuinely observed, whether the questions and answers were recorded faithfully, and whether the confession was retracted, since retraction does not destroy a confession the court believes was voluntary and true, though it sets the court looking for corroboration in prudence. An extra-judicial confession is everything else: the statement made to a private person, a relative, an employer, a fellow villager, a doctor, which the prosecution later seeks to prove through that person’s testimony. It clears Sections 25 and 26 where the maker was not in custody and the hearer was not a police officer, and it is judged under Section 24’s inducement screen like any other confession. Its weakness is structural rather than legal: it depends entirely on the memory and fidelity of the person reporting words spoken in private, often long before, frequently in a language requiring translation, and the courts therefore treat it as a weak species of evidence that must be proved with the precision expected of any other fact and accepted only where the reporting witness is credible and the words ascribed are unambiguous. The Supreme Court’s treatment in State of Rajasthan v Raja Ram in 2003 states the working position: an extra-judicial confession, if voluntary and true and made in a fit state of mind, can be relied upon, but the court must be satisfied that the witness is unbiased and reliable, and in practice courts look for corroboration from the surrounding circumstances before acting on such a statement alone. The two channels together show the confession law’s true shape: it is not a prohibition on confessions but a system of distrust management, closing the channel where coercion is hardest to detect, fortifying the channel where a neutral officer records, and discounting the channel where memory and motive are hardest to test.

Statements by the Dead: Section 32 and the Dying Declaration

Sections 32 and 33 govern statements made by persons who cannot be called as witnesses, because they are dead, cannot be found, have become incapable of giving evidence, or cannot be produced without an unreasonable amount of delay or expense. Section 32 declares eight classes of such statements relevant, each a controlled breach of the rule against hearsay, and the first of them is the most litigated statement in Indian criminal law: the dying declaration. The statutory text is broader than its English ancestor in two respects that decide real cases. The statement is relevant when it states the cause of the maker’s death or any of the circumstances of the transaction which resulted in the death, in a case in which the cause of that person’s death comes into question; and the section adds, in terms, that such statements are relevant whether the person who made them was, or was not, at the time of making them under expectation of death. English law, which insists on a settled and hopeless expectation of imminent death as the solemnity that substitutes for the oath, is therefore not the Indian rule, and a statement made by a victim who survived for days, or who did not know death was coming, can still be a dying declaration under the Act if it speaks to the cause or the circumstances of the transaction that in fact resulted in death.

The other clauses of Section 32 run the same logic across quieter terrain: statements made in the ordinary course of business, statements against the pecuniary or proprietary interest of the maker, statements giving the maker’s opinion as to public rights, customs and matters of general interest, statements about the existence of relationships by persons with special means of knowledge, and statements in documents about pedigree and family affairs. Section 33 performs the parallel service for evidence already given in a judicial proceeding: testimony from an earlier proceeding between the same parties may be read in a later one where the witness is dead or unavailable, provided the adverse party had the right and opportunity to cross-examine, the proceedings were between the same parties or their representatives, and the questions in issue were substantially the same. Both sections are governed by the same design instinct that runs through the Act: hearsay is excluded not because it is always worthless but because it is untested, and where the maker is beyond testing and the subject matter is one on which his statement carries structural guarantees of reliability, the Act lets the statement through the gate and leaves its worth to appreciation.

For the dying declaration, appreciation has been disciplined by two decisions that every treatment of the subject must carry. In Khushal Rao v State of Bombay, decided in 1957 and reported in 1958, the Supreme Court reviewed the authorities and laid down the propositions that still structure the inquiry: it cannot be laid down as an absolute rule of law that a dying declaration cannot form the sole basis of conviction unless corroborated; each case must be determined on its own facts, keeping in view the circumstances in which the declaration was made; a dying declaration is not a weaker kind of evidence than other evidence and stands on the same footing, to be judged by the ordinary principles governing the weighing of evidence; a declaration recorded by a competent Magistrate, in the form of questions and answers and as nearly as possible in the maker’s own words, stands on a much higher footing than one depending on oral testimony, which suffers from the infirmities of human memory and character; and the court testing reliability must keep in view the declarant’s opportunity for observation, including whether there was sufficient light, whether the capacity to remember was impaired, whether multiple declarations were consistent, and whether the statement was made at the earliest opportunity free of tutoring by interested parties. The same judgment states the corollary without flinching: where the court finds the declaration true and voluntary, no question of further corroboration arises, and where it finds infirmity, the declaration alone cannot convict.

The second decision resolves the recurring procedural skirmish over medical certification. In Laxman v State of Maharashtra in 2002, a Constitution Bench of five judges held that the fitness of the declarant’s state of mind is a matter the court must be satisfied about, and that medical opinion is the normal means of that satisfaction, but that a doctor’s certification of fitness is essentially a rule of caution rather than a condition of admissibility: where the person recording the declaration, or eyewitnesses, testify that the deceased was in a fit and conscious state to make it, the absence of a doctor’s certificate does not by itself make the declaration unacceptable, and medical opinion cannot prevail over the direct testimony of the recording witness on the point. Read together, the two cases mark out the mature position: the dying declaration is admitted on the statutory text without an expectation-of-death condition, screened at appreciation with close scrutiny because the accused never had the chance to cross-examine, and capable, when it survives that scrutiny, of carrying a conviction entirely on its own. The debate that persists is a value debate conducted inside those fixed points: whether a class of evidence immune from cross-examination should ever suffice alone, against which the answer the case law gives is that the alternative, silencing the victim of a fatal attack in every case where no other witness exists, would hand a class of killings an evidentiary immunity no legal system in this tradition has been willing to grant.

The Second Gate in Operation: Proof Under Part II

Part II begins by clearing the ground. Some facts need no proof at all: facts of which the court must take judicial notice under Section 57, a listed class running from the laws in force in India and the names of the members and officers of Parliament and legislatures to the divisions of time, the geographical divisions of the world, public festivals and the rules of the road on land and at sea, and facts admitted by the parties or their agents under Section 58, since a fact a party concedes is a fact the opposing party is relieved of proving. Everything else must be proved by evidence, and Part II then divides the labour by kind. Oral evidence is governed by two short provisions of large effect. Section 59, as recast by the Information Technology Act, 2000, provides that all facts, except the contents of documents or electronic records, may be proved by oral evidence. Section 60 then insists that oral evidence must in all cases be direct: the witness who says he saw the event must have seen it, the witness who says he heard the statement must have heard it, and a witness relaying what another person perceived is giving hearsay, which Section 60’s insistence on directness shuts out. The hearsay rule in Indian law is not a separately stated prohibition; it is the consequence of Section 60 read with the closed list of relevant facts, and Sections 32 and 33 are intelligible only as the Act’s deliberate, enumerated exceptions to it.

Documentary evidence occupies Sections 61 to 100, and its organising distinction is between a document itself and everything else that reports its contents. Section 61 requires the contents of a document to be proved by primary or secondary evidence. Section 62 defines primary evidence as the document itself produced for the inspection of the court, with explanations that handle documents executed in several parts, counterparts, and uniform processes such as printing, lithography and photography. Section 63 defines secondary evidence in five classes: certified copies given under the Act, copies made from the original by mechanical processes that ensure accuracy and copies compared with such copies, counterparts as against the parties who did not execute them, and an oral account of the contents by a person who has seen the document. Section 64 then states the default rule: documents must be proved by primary evidence except in the cases Section 65 mentions. Section 65 is the gate through which every copy must pass, and it opens on seven grounds keyed to the whereabouts and condition of the original: the original is in the possession of the adversary or a person out of reach or not subject to the court’s process; the adversary has admitted its existence and contents in writing; the original has been destroyed or lost, or the party cannot produce it in reasonable time for a reason not arising from his own default or neglect; the original is not easily movable; the original is a public document; the original is one of which a certified copy is permitted by law; and the case concerns numerous accounts that cannot conveniently be examined. Section 66 adds the discipline of notice: secondary evidence of a document in the adversary’s possession generally requires notice to produce it first, so that the original gets its chance to appear. Around this core sit the provisions on proving execution, signature and handwriting in Section 67, the attestation rules for documents the law requires to be attested in Sections 68 to 72, and the public document regime of Sections 74 to 78, under which certified copies of public documents, the records of public offices, courts and legislatures, may be produced to prove their contents.

Chapter VI of Part II, Sections 91 to 100, states the exclusion rule for documents that lay readers rarely expect and practitioners use constantly. Where the terms of a contract, grant or disposition of property have been reduced to the form of a document, Section 91 requires the document itself, or secondary evidence of its contents where that is admissible, as the only proof of those terms, and Section 92 then forbids the parties to contradict, vary, add to or subtract from those terms by oral agreement, subject to provisos that admit evidence of facts invalidating the document, of a separate oral agreement on a matter the document is silent about where it is not inconsistent with its terms, of conditions precedent, and of usage and custom that annex incidents the document does not mention. The architecture is the same three-gate discipline applied inside a single transaction: the parties chose a written form, and the Act holds them to it, because a rule that allowed every written contract to be rewritten by oral testimony would make the written form pointless. Sections 93 to 100 work out the adjacent rules for ambiguous language, documents applying partly to one set of facts and partly to another, and evidence explaining latent ambiguities.

The closing provisions of Part II equip the court with presumptions about documents, devices that allocate the risk of missing proof. Section 79 presumes the genuineness of certified copies; Section 80 presumes the accuracy of documents produced as the record of evidence given in judicial proceedings; Section 81 covers gazettes and newspapers; Section 83 covers maps and plans made by governmental authority; Section 85 presumes due execution of powers of attorney executed before specified officers; Sections 86 to 89 continue the pattern across certified copies of foreign judicial records, collections of laws and reports of decisions, and documents withheld after notice to produce; and Section 90, the provision of daily importance in property litigation, permits the court to presume that a document thirty years old, produced from custody the court considers proper, was duly signed, written and executed by the persons it purports to come from. These are presumptions under Section 4 in its middle register, matters the court may presume, regarding a fact as proved unless and until it is disproved, or calling for proof of it. Their function within the filter is precise: they do not decide what is true, they decide who loses if the truth about an old piece of paper cannot be reconstructed, and they are the first appearance in the Act of the burden-shifting machinery that Part III develops into a system.

Proving Execution: Attestation, Signatures, and the Documentary Corners

A document that passes the production rules must still be proved as what it purports to be, and Chapter V of Part II devotes its second half to the proof of execution. Section 67 states the general requirement: where a document is alleged to be signed or to have been written wholly or partly by any person, the signature or the handwriting of the part alleged to be his must be proved to be in his handwriting. The modes of proof are assembled from across the Act: the testimony of the writer himself, of a person who saw the document written or signed, of a person acquainted with the handwriting under Section 47’s conditions, of an expert under Section 45, and of the court comparing the disputed writing with a writing admitted or proved under Section 73. For documents that the law requires to be attested, a will being the case that generates the litigation, Section 68 adds a condition with teeth: the document shall not be used as evidence until one attesting witness at least has been called for the purpose of proving its execution, if there is an attesting witness alive, subject to the court’s process and capable of giving evidence. The registered document carries a statutory softening for instruments other than wills: where a document required by law to be attested has been registered under the registration law in force, the calling of an attesting witness is not necessary unless its execution is specifically denied, a proviso that rescues countless registered deeds from an impossible search for witnesses decades later. Section 69 then provides the route where no attesting witness can be found, which includes the case of witnesses dead or beyond the court’s reach: the party must prove that the attestation of at least one attesting witness is in his handwriting and that the signature of the executant is in the executant’s handwriting. Section 70 makes the executant’s own admission of execution sufficient proof against him, and Section 71 permits execution to be proved by other evidence where an attesting witness denies the execution or does not recollect it, the provision that rescues documents from the attesting witness who turns hostile in the technical rather than the colloquial sense. The cumulative effect is a proof ladder: call the attesting witness where one can be called, descend to handwriting proof where one cannot, and use the opponent’s own admission wherever it exists, and the litigation over wills in particular is largely the story of parties climbing and contesting that ladder.

Two smaller provisions of Part I and Part II do recurring work in commercial litigation and belong in a complete account of the proof gate. Section 34 makes entries in books of account regularly kept in the course of business relevant whenever they refer to a matter the court must inquire into, but adds a discipline unusual in the Act: such statements shall not alone be sufficient evidence to charge any person with liability. The ledger entry is therefore a door-opener and not a verdict; it must be supported by evidence of the underlying transaction, and the courts have built the practice of account-proof on the distinction. Section 35 performs the same service for entries in public and official records, registers and reports made by a public servant in the discharge of his duty, or by any other person in performance of a duty specially enjoined by law, making the entry itself a relevant fact. Both provisions show the Act handling the everyday reality that commerce and administration generate paper in bulk: the entries come in, because a rule excluding them would exclude the ordinary record of economic life, and they come in with a built-in caution about how far they can carry a case by themselves. A party proving a running account therefore works the whole sequence this article has described: the entries are relevant under Section 34, the books must be produced as primary evidence of their contents or their absence explained under Section 65, and the transactions the entries record require independent proof before liability can be charged on them.

Electronic Records: The Certificate That Decided Thousands of Cases

The amendment that most altered the working life of this nineteenth-century Act arrived with the Information Technology Act, 2000, which recast the definition of evidence in Section 3 to include electronic records within documentary evidence, barred oral admissions as to the contents of electronic records unless the genuineness of the record is in question under the new Section 22A, and inserted Sections 65A and 65B as a self-contained code for proving electronic records. The reason for a separate code is a property of the material. A paper document exists in one authoritative instance and copies degrade by comparison; an electronic record exists as information that can be reproduced perfectly, altered without trace, stored in one device and displayed on another, and detached from any original a court could sensibly demand. The general documentary rules, built on the distinction between the document itself and reports of its contents, do not map cleanly onto a world in which the printout, the screenshot and the forwarded file are all, in an obvious sense, copies. Parliament’s answer, following the approach of the Information Technology Act itself, was to define the conditions on which computer output would be treated as a document and admitted without proof of the original. The reader tracing the parent statute’s own design will find it in the guide to the Information Technology Act 2000; what matters here is what the amendment did inside the evidence filter.

Section 65A states that the contents of electronic records may be proved in accordance with Section 65B, and Section 65B does the work. It deems computer output, meaning information printed on paper or stored, recorded or copied in optical or magnetic media produced by a computer, to be a document admissible in evidence without further proof or production of the original, if the conditions of the section are satisfied. The conditions in Section 65B(2) police the regularity of the machine process: the computer output was produced by a computer used regularly to store or process information for activities regularly carried on; information of the kind contained in the record was regularly fed into the computer in the ordinary course of those activities; the computer was operating properly throughout the material period, or any malfunction did not affect the record or its accuracy; and the information in the output reproduces or is derived from the information so fed. Section 65B(4) then requires the certificate: a document identifying the electronic record and describing the manner of its production, giving particulars of the device involved, dealing with the conditions, and signed by a person occupying a responsible official position in relation to the operation of the relevant device or the management of the relevant activities. Two distinctions in the scheme decide most disputes. The first is between the original device and its output: where the device itself, the phone or the computer, is produced in court and its owner testifies, the record is proved as primary evidence and no certificate is needed, as the Supreme Court later confirmed. The second is between output tendered as secondary evidence, a printout, a copy, a clone, where the certificate is the statutory price of admission.

The certificate requirement generated a three-act litigation sequence that is the clearest available illustration of how a single sub-section can govern the fate of thousands of cases. The first position came in State (NCT of Delhi) v Navjot Sandhu in 2005, the Parliament attack case, where the Supreme Court treated Section 65B as one available route among others and permitted secondary evidence of electronic records to be given under the general provisions of Sections 63 and 65 even without strict compliance, a permissive reading that allowed call records supported by oral testimony to pass the gate. The second act is Anvar P V v P K Basheer in 2014. Anvar arose from an election petition in which songs and speeches alleged to be corrupt practices were circulated on compact discs, and the discs were sought to be proved without certificates. A three-judge bench held that Section 65B is a special provision and a complete code on electronic evidence: where a party relies on an electronic record as secondary evidence, compliance with Section 65B, including the certificate under Section 65B(4), is a condition precedent to admissibility, the general provisions for documents cannot be used to bypass it, and the permissive reading in Navjot Sandhu was not good law. The third act began when a two-judge bench in Shafhi Mohammad v State of Himachal Pradesh in 2018, confronting the practical problem of a party who did not control the device, treated the certificate requirement as procedural and held that it could be relaxed in the interests of justice where the party was not in possession of the device from which the record was produced.

The sequence closed with Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal in 2020, where a three-judge bench on a reference restored Anvar and overruled both Shafhi Mohammad and the intervening decision in Tomaso Bruno v State of Uttar Pradesh that had followed the permissive line. The holding has several limbs, and stating it as merely “certificate mandatory” discards half of what the judgment decided. The certificate is a condition precedent for secondary evidence of electronic records, and the requirement cannot be dispensed with on grounds of convenience or inability simpliciter. But the requirement fastens on the stage of admissibility and not on the moment of filing: a party that has been unable to produce the certificate with the record may produce it at a later stage of the proceedings, and where the person in charge of the device will not issue it, the party’s remedy is to apply to the court, which can summon that person and order production, using the court’s ordinary coercive powers over witnesses and documents; a party that never takes the steps the judgment lays out cannot claim the benefit of impossibility. Objections to the mode of proof, as distinct from objections to relevance, must be taken when the evidence is tendered and not raised for the first time in appeal. And the Court issued directions to cellular and internet service providers to maintain records, including call detail records, segregated and securely, for the period the law stipulates, precisely because the certificate regime is workable only if the underlying records survive until the trial. The position that emerges is strict in substance and structured in procedure: the gate is real, the price is the certificate, and the route for a party genuinely locked out of the device runs through the court rather than around the gate.

The practical failures this regime produces are repetitive enough that trial teams build their hearing preparation around them. The screenshot offered as though it were self-proving fails, because a screenshot is computer output and carries the certificate question with it. The chat export filed by an individual who no longer holds the phone fails at the signature line, because the certificate must come from the person in the responsible position in relation to the device, and where that person is an opponent or a platform, the Arjun Panditrao route of a court application is the mechanism the law supplies. The forensic image proved through the Examiner of Electronic Evidence, the statutory expert whose opinion Section 45A, inserted by the Information Technology (Amendment) Act, 2008, makes a relevant fact, raises the separate question of genuineness that arises only after admissibility is secured, and conflating the two questions, as advocates and commentators persistently do, repeats in miniature the error this article began with, the collapse of separate gates into one. It should be said plainly, because the position described in this article is general information about a statutory scheme and not advice on any individual case, that certificate defects are among the most unforgiving in the law of evidence: a record excluded at the gate cannot be argued at appreciation, and the time to repair the defect is during the trial and not after it. Practitioners assembling a record for a hearing, paginating exhibits and arranging the sequence in which documents will be taken up, often work through the bundle as a set of slides before the hearing begins, and a reader handling such material on a screen can open and review presentation files directly in the browser through the ReportMedic presentation viewer rather than installing local software for the purpose.

After Admissibility: Genuineness, Hash Values, and the Camera That Was Not Produced

Securing the admission of an electronic record settles only its right to be in the room, and the litigation then moves to the question the certificate does not answer: is the record genuine, and does it show what the party says it shows? The certificate under Section 65B speaks to the regularity of the computer process and the identity of the record as an output of that process. It does not vouch for the truth of the record’s contents, any more than the proof of a letter’s execution vouches for the truth of what the letter asserts, and it does not exclude manipulation that occurred before the output was generated. Parties therefore fight the genuineness battle with the tools the Act gives for documents generally: the forensic examination of the device and the storage media, the opinion of the Examiner of Electronic Evidence under Section 45A where the information’s character is in issue, the metadata carried within files, and hash values, the algorithmic fingerprints whose match between two copies demonstrates that the copies are bit-for-bit identical and whose mismatch demonstrates alteration after the reference value was taken. A hash value proves identity of content; it does not prove provenance, and the litigated questions in practice are which file was hashed, when, and by whom, questions answered by witnesses and process evidence like any other foundational fact. The sequence matters in argument: genuineness objections raised against a record that has not yet passed the certificate gate are premature, and authenticity evidence led for a record that fails the gate is wasted, which is why the discipline of separating admissibility from appreciation, running through this entire article, is nowhere more operationally important than here.

The camera cases show the other side of the electronic coin, the record that never reaches the filter at all. Closed-circuit footage is electronic evidence, and where it exists it is often the best evidence of an event, which is why the Supreme Court in Tomaso Bruno v State of Uttar Pradesh in 2015, a case about a death in a hotel, treated the investigation’s failure to collect and produce the available camera footage as a serious infirmity in the prosecution case, reasoning that the best available evidence had been withheld from the court. The doctrinal hook for that reasoning sits in Section 114: illustration (g) permits the court to presume that evidence which could be and is not produced would, if produced, be unfavourable to the person who withholds it. Read with Tomaso Bruno, the illustration converts the prosecution’s duty of fair investigation into an evidentiary consequence: footage that existed, was obtainable, and is absent will be reasoned about adversely, and an accused who can point to a missing camera record has an argument that operates entirely within the Act’s own machinery, without any need to dispute what the surviving evidence shows. That Tomaso Bruno’s permissive position on certificates was later overruled while its insistence on producing the best evidence stands unshaken is a neat summary of the whole electronic evidence story: the courts relaxed nothing on the duty to produce, and argued for a decade only about the paperwork that accompanies production.

Burden of Proof and the Presumption Machinery: Sections 101 to 114A

Part III opens with the question every other part presupposes: who must prove the facts that the filter has admitted in principle? Section 101 states the master rule in terms of consequence: whoever desires a court to give judgment as to a legal right or liability dependent on facts he asserts must prove that those facts exist, and a person is said to bear the burden of proof when he is bound to prove a fact. Section 102 locates the burden on the party who would fail if no evidence at all were given on either side, and Section 103 places the burden as to any particular fact on the person who wishes the court to believe in its existence, unless a law provides otherwise. Section 104 completes the set by putting the burden of proving a fact that must be proved before other evidence becomes admissible on the party who wants to give that evidence, the foundation rule that Section 136’s conditional admissibility relies upon. Two consequences follow that organise the whole subject. The burden on the whole case, in a criminal trial the prosecution’s burden to prove guilt, never shifts; the burden of proving individual facts moves constantly as evidence accumulates, and the Act’s text supports both propositions at once because its sections are addressed to facts and not to cases. And the standard is held within the Section 3 definition of proved and disproved: in civil matters the prudent person acts on a preponderance of probabilities, while in criminal matters the calibration of that same definition demands proof beyond reasonable doubt, a distinction the courts derive from the gravity of what the prudent person is being asked to act upon.

Section 105 places a specific and frequently misunderstood burden on the accused: where an accused person claims that his case falls within one of the general exceptions in the Penal Code, or within a special exception or proviso in the law defining the offence, the burden of proving the existence of the exculpating circumstances lies on him, and the court shall presume the absence of such circumstances. The accused’s burden under this section is not the prosecution’s burden, and the case law consistently measures it by the civil standard of preponderance of probabilities, so that an accused may discharge it by showing his version of self-defence or grave and sudden provocation is the more probable account, without disturbing the prosecution’s obligation to prove the offence itself beyond reasonable doubt. Section 106 then provides that when a fact is especially within the knowledge of any person, the burden of proving that fact is upon him. The section is indispensable and dangerous in equal measure, and the Supreme Court in Shambhu Nath Mehra v State of Ajmer in 1956 fixed its limits in terms that are still the controlling statement: Section 106 is an exception to the general rule of Section 101, it cannot be used to relieve the prosecution of its duty to establish the foundational facts of its case, and it applies only where the fact is of a kind that is impossible or disproportionately difficult for the prosecution to prove and comparatively simple for the accused to explain. The illustration that governs practice is the railway passenger charged with travelling without a ticket, who knows whether he paid; the counter-illustration, the prosecution inviting the court to convict because the accused has not explained how the deceased came to die in his company, is the misuse the judgment forbids unless the prosecution has first proved the setting that makes the fact specially the accused’s to explain.

Sections 107 to 114A then distribute presumptions across the recurring situations where proof is structurally difficult, using all three registers Section 4 supplies. The court shall presume, regarding the fact as proved unless and until it is disproved, in the two modern presumption provisions this article must state carefully. Section 113A, inserted by the Criminal Law (Second Amendment) Act, 1983, permits the court to presume that a married woman’s suicide was abetted by her husband or his relatives where the suicide occurred within seven years of the marriage and the husband or relatives subjected her to cruelty; note that the register is “may presume” in the text of the discretion the section gives, and the court retains the choice the words confer. Section 113B, inserted by the Dowry Prohibition (Amendment) Act, 1986, is framed in the mandatory register: where the question is whether a person has caused the dowry death of a woman, and it is shown that soon before her death she was subjected by that person to cruelty or harassment for or in connection with a demand for dowry, the court shall presume that the person caused the death. Section 114A, inserted by the Criminal Law (Amendment) Act, 2013, directs that in a prosecution for specified categories of rape, where sexual intercourse by the accused is proved and the woman states in her evidence before the court that she did not consent, the court shall presume that she did not consent. Each of these provisions performs the same operation in a domain where the decisive facts occur in private and proof is chronically scarce: the prosecution must still prove the foundation facts, cruelty, the dowry demand close in time to the death, the act of intercourse, and the presumption then supplies the element that direct evidence rarely reaches. Misdescribing any of them, as commentary routinely does, as reversing the burden of proof wholesale is an error the text does not support; the burden on the foundation never moves.

Section 114, the widest provision in the chapter, empowers the court to presume the existence of any fact it thinks likely to have happened, regard being had to the common course of natural events, human conduct, and public and private business, in relation to the facts of the particular case, and its illustrations function as a pocket code of forensic common sense: that a man in possession of stolen goods soon after the theft is either the thief or received them knowing them to be stolen; that an accomplice is unworthy of credit unless corroborated in material particulars; that a document produced is genuine if it comes from proper custody; that evidence which could be and is not produced would, if produced, be unfavourable to the party withholding it. The last of these, illustration (g), is the engine of adverse inference in Indian litigation, and it is a presumption in the discretionary register, a tool the court may deploy where the withholding is unexplained. Beside these sit the older fixed presumptions: Section 107 and Section 108 allocate the burden where a person is shown to have been alive within thirty years or not heard of for seven years by those who would naturally have heard of him; Section 110 treats possession as evidence of ownership and puts the burden on the challenger; Section 111 puts the burden of proving good faith on the party in a position of active confidence; Section 112 makes birth during marriage conclusive proof of legitimacy unless the parties are shown to have had no access to each other, one of the rare uses of the conclusive register in the Act, which forbids evidence in rebuttal altogether; and Section 111A, inserted for cases arising in areas declared disturbed, presumes that a person prosecuted for specified offences against the State in such an area, in whose possession arms, ammunition or explosives were found in circumstances raising a reasonable suspicion of the offence, committed that offence. The chapter as a whole is the Act’s answer to the brute fact that litigation is decided under uncertainty: where proof runs out, the statute decides in advance who loses, and on what showing the loss can be averted.

Estoppel, Witnesses, and the Conduct of Examination

Between the presumptions and the witnesses sits Section 115, the estoppel provision, which belongs to a different logical family from everything around it: it does not help prove a fact, it forbids a party from denying one. When a person has, by declaration, act or omission, intentionally caused or permitted another person to believe a thing to be true and to act upon that belief, neither the first person nor his representative in interest is allowed, in any suit or proceeding between them, to deny the truth of that thing. The doctrine’s illustration is the tenant and the landlord: a tenant who entered possession under a landlord is not permitted, while the tenancy continues, to deny that the landlord had title at its beginning. Estoppel under the Act is a rule of evidence and not a cause of action; it seals a party’s mouth about a fact his own conduct created reliance upon, and it operates only between the parties to the reliance, which is why it sits in Part III as a rule about the effect of evidence rather than among the relevancy provisions of Part I.

Chapter IX of Part III turns to witnesses, the human half of the definition of evidence. Section 118 states the competency rule in the Act’s characteristic negative form: all persons are competent to testify unless the court considers that they are prevented from understanding the questions put to them, or from giving rational answers to those questions, by tender years, extreme old age, disease of body or mind, or any other cause of the same kind. Competency is the default and incapacity the exception to be found by the court, so a child witness is not incompetent by reason of being a child; the inquiry is whether the particular child understands the questions and answers them rationally, a preliminary examination the courts conduct with children and with witnesses whose capacity is in doubt. The section’s explanation preserves the same functional test for a person with mental illness, who is not incompetent by reason of the illness unless prevented by it from understanding and answering. Section 119 deals with witnesses unable to speak, who may give evidence in any manner in which it can be made intelligible, by writing or by signs made in open court, evidence so given being deemed oral evidence. Section 120 makes the parties to a civil suit, and the spouses of parties, competent witnesses, and in criminal proceedings makes the accused’s spouse a competent witness as well, while leaving the accused’s own position, the right not to be compelled to testify, to the criminal procedure and constitutional law that govern it.

Sections 121 to 129 then fence off the communications the law will not reach even though relevant, the true privileges of the Act, and their placement inside the witness chapter reflects that they operate as refusals at the point of production. A judge or Magistrate cannot be compelled to answer questions about his own conduct in court or matters that came to his knowledge in court in that capacity, Section 121, save on the special order of a court to which he is subordinate. Communications between spouses during marriage are protected by Section 122: neither spouse can be compelled to disclose them, nor permitted to disclose them without the other’s consent, except in proceedings between the spouses or where one is prosecuted for a crime against the other. Section 123 protects unpublished official records relating to affairs of state, producible only with the permission of the head of the department; Section 124 protects official communications made in confidence where the public interest would suffer by disclosure; Section 125 shields Magistrates and police and revenue officers from being compelled to name the source of information about offences. The professional communications regime of Sections 126 to 129 protects communications between a client and his barrister, attorney, pleader or vakil made in the course of and for the purpose of the employment, together with the contents of documents the adviser has seen in that course, subject to exceptions that mark the privilege’s moral limits: communications made in furtherance of an illegal purpose are not protected, and facts the adviser observes showing that a crime or fraud has been committed since the employment began fall outside it. The privilege belongs to the client, survives the end of the employment, and can be waived only by the client’s express consent.

The witness who does testify is then placed under a regime of answerability that surprises readers who expect a general right of silence in the box. Section 132 provides that a witness shall not be excused from answering a relevant question on the ground that the answer will criminate him or expose him to penalty or forfeiture, and balances the compulsion with a proviso of real value: no such answer shall subject him to any arrest or prosecution, or be proved against him in any criminal proceeding, except a prosecution for giving false evidence by that answer. Section 133 makes an accomplice a competent witness against an accused person and provides that a conviction is not illegal merely because it rests on the uncorroborated testimony of an accomplice, a statutory permission that Section 114’s illustration (b), the presumption that an accomplice is unworthy of credit unless corroborated in material particulars, immediately disciplines into the working rule courts actually follow. Section 134 then states the principle this series of provisions builds toward: no particular number of witnesses is required for the proof of any fact. Quantity is irrelevant to the filter; a single witness believed is proof, and a dozen witnesses disbelieved are not, which is why the appreciation stage and not the admission stage carries the real weight of the Act.

Chapter X governs the examination itself. Section 137 defines examination-in-chief, cross-examination and re-examination, and Section 138 fixes their order and the scope of the second and third rounds: cross-examination need not be confined to the facts the witness testified to in chief, and re-examination is directed to the explanation of matters referred to in cross, with new matter admissible only by leave of the court and then open to further cross. Section 141 defines a leading question as one suggesting the answer the questioner wishes or expects to receive, Section 142 forbids leading questions in examination-in-chief and re-examination where the other party objects, and Section 143 permits them freely in cross-examination, the scheme reflecting the assumption that a party’s own witness is friendly and the opponent’s examination is the testing engine of the trial. Section 145 controls the contradiction of a witness by his previous written statement, requiring the relevant parts to be put to him before the writing can be used to contradict him; Section 146 opens cross-examination to questions testing veracity, discovering who the witness is and his position in life, and shaking his credit, subject to the court’s control in Sections 148 to 152 over questions that are indecent, scandalous or intended to insult or annoy; Section 155 lists the modes of impeaching a witness’s credit, including proof of bribery or corrupt inducement and former inconsistent statements; Sections 159 to 161 govern the refreshing of memory from writings made contemporaneously with the events; and Section 165 arms the judge with the power to put any question, in any form, at any time, to any witness or party, about any fact relevant or irrelevant, and to order the production of any document or thing, while confining the judgment that follows to facts that are relevant and duly proved. Section 154, placed among these provisions, governs the witness who turns: the court may in its discretion permit the party who called a witness to put to him any questions that could be put in cross-examination by the adverse party. The Act never uses the label hostile witness, which is the profession’s shorthand for the situation, and the Supreme Court in Sat Paul v Delhi Administration in 1976 settled the two errors the shorthand invites: a witness granted the Section 154 permission is not thereby branded a liar whose testimony must be discarded, and the evidence of such a witness remains evidence, capable of being relied on by either side to the extent the court finds it trustworthy after the testing the permission enables.

What happens when inadmissible evidence is admitted without objection?

Section 167 provides that improper admission or rejection of evidence is not, of itself, a ground for a new trial or reversal, where the appellate court finds that, independently of the evidence objected to, sufficient evidence justified the decision, or that the rejected evidence, if received, ought not to have varied it.

Gate errors matter, in other words, only where they can be shown to have mattered, and the provision makes appellate relief turn on materiality rather than on error alone.

Objections, Marking, and the Life of the Record

Between the gates and the trial stands a small procedural economy that the Act presupposes and that practitioners ignore at the cost of appeals. Evidence is tendered item by item; a document is shown to the witness or the court, formally offered, and, if received, marked as an exhibit, while a document refused remains on the file marked for identification or simply recorded as tendered and rejected. The marking is not a ruling on proof. An exhibit number records that the document was received into the record, and the questions whether it was duly proved, what it establishes, and whether the mode of proof was lawful remain open for argument at the end of the trial and on appeal, a distinction between reception and proof that the three-gate analysis predicts and that the case law on objections enforces. Objections themselves divide by kind, and the division decides when they may be raised. An objection to relevancy or to admissibility in the strict sense, that the item is of a kind the law excludes, goes to the root and can in principle be entertained even on appeal, because evidence that should never have been before the court cannot be legitimised by silence. An objection to the mode of proof, that the item is admissible in kind but has been proved in an irregular manner, a copy received without the Section 65 groundwork or an electronic record received without its certificate, must be taken when the evidence is tendered, because the defect is curable at that moment and the party tendering is entitled to the chance to cure it; taken for the first time on appeal, it will generally fail, as the Supreme Court reiterated in the electronic records line when it channelled certificate objections into the trial. Section 167 then governs the appellate consequences of error at the gate, as the discussion of that provision above shows, reversing only where the wrongly admitted matter could have affected the decision or the wrongly rejected matter should have changed it.

The record this process produces is the only thing an appellate court ever sees, and its quality is decided by habits the Act rewards. A ruling on an objection, given at the moment and recorded with its reason in brief, protects both sides: the receiving side has its item, the objecting side has its point preserved, and the appellate court can locate the error if there was one. A trial in which objections are waived through as a matter of routine, documents marked without proof, and certificates untendered, produces a record in which the gates exist on paper and nowhere else, and the pattern recurs in appellate judgments setting aside convictions built on unproved documents. None of this machinery appears in the Act’s text as a code of trial management, because the Act is a filter and not a manual; but the filter’s provisions are written on the assumption that someone at the hearing is operating it, item by item, and the judge’s powers under Section 136 and Section 165 are the statutory handles by which that operation happens.

Character Evidence and the 2013 Corrections

Part I closed with the character provisions, and their working life deserves a fuller statement than their position in the relevancy catalogue suggests, because character is where the Act’s protective asymmetries meet the realities of cross-examination. The civil rule of Section 52 excludes character offered to prove that a party’s conduct was probable or improbable, on the ground that civil liability turns on acts and not dispositions; the exceptions, character appearing from facts otherwise relevant and character affecting damages under Section 55, mark the points at which disposition is itself part of the inquiry. The criminal scheme of Sections 53 and 54 is deliberately uneven. The accused may prove his previous good character, and the fact is relevant, because a person of good character is, in the Act’s estimation, less likely to have committed the act charged and entitled to have the improbability weighed. The prosecution may not prove the accused’s bad character as part of its case in chief, and may do so only in reply where the accused has asserted good character, or where the accused’s character is itself a fact in issue, or where bad character becomes relevant through the evidence the defence has led. Behind those sections stands Section 54’s explanation protecting previous convictions from casual deployment, and the scheme as a whole embodies a judgment about fair trial: the prosecution must prove the act, not assemble a disposition and invite the court to reason backwards from it.

Cross-examination is where character returns by the side door, and Sections 146 and 155 control that traffic. A witness may be asked questions in cross-examination to test his veracity, to discover who he is and his position in life, and to shake his credit by injuring his character, even where the answer might expose him to a penalty, subject to the court’s control and to Section 132’s protection of compelled answers. The use of that licence against the complainant in a sexual offence trial was historically its most criticised application: Section 155 originally permitted, in its clause (4), the credit of a prosecutrix in a rape prosecution to be impeached by evidence that she was of generally immoral character, a provision that converted the trial of the offence into a trial of the complainant’s past. The Criminal Law (Amendment) Act, 2013 removed that clause from Section 155 and inserted Section 53A, providing that in a prosecution for specified sexual offences, evidence of the character of the victim or of the victim’s previous sexual experience with any person shall not be relevant to the issue of consent or the quality of consent. The amendment arc of 2013, treated in its own article in this series, is therefore legible inside the evidence code as a change in what the filter will let through on the subject of character: the accused’s shield of good character remains, the complainant’s past is withdrawn from the contest over consent, and the court’s control of credit questions under Sections 148 to 152 supplies the residual discipline for the cross-examination that remains.

Appreciation: The Fourth Operation the Text Barely Names

The Act regulates the first three operations in detail and the fourth, appreciation, mostly by implication, because appreciation is judgement itself and resists codification. Yet the case law of appreciation is the part of evidence law that decides outcomes, and three of its frameworks are stable enough to state as doctrine. The first concerns the single witness. Since the Act requires no particular number of witnesses, the question is never how many but of what quality, and the Supreme Court in Vadivelu Thevar v State of Madras in 1957 sorted witnesses into the three categories that courts still use: the wholly reliable, whose testimony may be acted on without hesitation; the wholly unreliable, whose testimony must be discarded; and the neither wholly reliable nor wholly unreliable, whose testimony requires corroboration in material particulars before it can be acted upon safely. The classification is applied witness by witness and often part by part, since a witness may be believed about one part of an account and disbelieved about another, and the discipline it imposes, find the category before finding the fact, is the closest the system comes to a method for appreciation.

The second framework governs circumstantial evidence, the chain of facts from which guilt is inferred where no witness saw the offence. The formulation in Hanumant v State of Madhya Pradesh in 1952 required that the circumstances be fully established and consistent only with the hypothesis of guilt, and the five principles restated by the Supreme Court in Sharad Birdhichand Sarda v State of Maharashtra in 1984 remain the standard recitation: the circumstances from which the conclusion of guilt is drawn should be fully established; the facts so established should be consistent only with the hypothesis of the guilt of the accused and not explainable on any other hypothesis except that the accused is guilty; the circumstances should be of a conclusive nature and tendency; they should exclude every possible hypothesis except the one to be proved; and there must be a chain of evidence so complete as not to leave any reasonable ground for a conclusion consistent with innocence, showing that in all human probability the act must have been done by the accused. The framework’s severity is deliberate: an inference is doing the work that perception does in direct evidence, and the five conditions are the price of letting it. The third framework is the accomplice practice already noted: Section 133 permits conviction on an accomplice’s uncorroborated testimony as a matter of law, Section 114 illustration (b) presumes the accomplice unworthy of credit unless corroborated in material particulars, and the courts have resolved the apparent tension by treating corroboration as a rule of practice so settled that it has hardened into a rule of law, requiring independent evidence connecting the accused with the offence and not merely confirming that an offence occurred.

Appreciation also governs the conversation between kinds of evidence when they disagree, and two of its standing rules recur often enough to belong in any account of the filter’s last stage. Medical and ocular evidence are not ranked in advance; the court asks whether the variance is one of degree, in which the eyewitness account of an event survives differences in the medical reconstruction of its details, or one of kind, in which the medical evidence rules out the eyewitness version altogether, in which case the version falls. And the testimony of an interested or related witness is not discounted for the relationship alone; the courts scrutinise it with care and act on it where it bears the impress of truth, since in the settings where much Indian crime occurs, the family and the village are the witnesses there are. These frameworks share a character worth naming at the end of the body of this article: they are none of them in the Act’s text, and all of them are built on the Act’s foundations, the definition of proved in Section 3, the burden provisions, the illustration structure of Section 114. The statute supplies the grammar; the courts, over a century and a half, supplied the syntax of judgement.

The Filter at Work: Four Items Through the Gates

The gates are easier to hold in mind against concrete material, and four recurring items show the sequence deciding real disputes. Take first the printed chat thread, the single most common electronic exhibit in contemporary litigation. At the relevancy gate the thread passes easily, since the messages are facts forming part of the same transaction or explaining the conduct in issue, under Sections 6 and 8. At the mode-of-proof gate the printout is computer output, secondary evidence of the record on the phone, and the certificate under Section 65B(4) is demanded, identifying the record, describing the manner of production, giving the particulars of the device, and signed by the person in the responsible position in relation to it. If the phone itself is produced and its owner testifies to the thread, the record comes as primary evidence and the certificate question falls away. Suppose the certificate is produced and the thread is admitted; nothing yet has been proved about the decisive point, the identity of the sender, because admissibility of the record and attribution of the messages are separate questions. Attribution is fought at appreciation, on the custody of the phone, the pattern of the exchanges, admissions elsewhere, and the presumption the court may draw under Section 114 from a party’s failure to produce the device or the records within his control. A litigant who treats the certificate as victory has confused the second gate with the fourth.

Take next the custodial statement that leads to a buried weapon, the standard Section 27 contest. The statement as a whole is a confession in custody and Sections 25 and 26 exclude it. The discovery changes the analysis for part of it: the fact that the weapon was found at the place the accused indicated is deposed to by the recovery witnesses, and so much of the information as relates distinctly to that discovery may be proved. The trial then becomes an exercise in the Kottaya discipline. The prosecution must show the fact was genuinely discovered in consequence of the information, so a weapon the police already knew of, or a place already searched, yields nothing; the admissible fragment is severed from the statement’s admissions about the offence itself, the words of past use staying out; and the court tests the recovery evidence, the memo of the statement, the independence of the recovery witnesses, the sealing and production of the weapon, as it tests any other contested fact. At appreciation, the discovery proves the accused’s knowledge of the place of concealment, and the inference from that knowledge to participation in the offence is a further step the court takes or declines on the whole record. Each stage has its own failure mode, and the cases in which Section 27 evidence collapses are almost always cases in which a stage was skipped: severance not performed, discovery not genuinely new, or the inference treated as automatic.

Take the thirty-five-year-old registered sale deed produced in a suit over title. Relevancy is direct, the deed being the operative document of the transaction in issue. At the proof gate, the original is produced, so primary evidence under Section 62 is before the court, and the question moves to execution: the deed being a document required by law to be attested, Sections 68 to 71 govern proof of execution through attesting witnesses, with the routes the sections provide where an attesting witness is dead, untraceable or denying. Section 90 then works on the residue of doubt about the paper itself: produced from proper custody after thirty years, the court may presume the signatures and handwriting and due execution and attestation. Had a party produced only a photocopy, the proof gate would have demanded the Section 65 groundwork instead, the absence or loss of the original accounted for, and had the opposing party relied on an oral account of what the deed said, Sections 91 and 92 would have shut it out, because the terms of a disposition reduced to a document must be proved by the document and cannot be varied by oral evidence. The example shows why the Act’s documentary chapters are the most heavily used in civil practice: property disputes are won at the second and third gates far more often than at the first.

Take finally the letter a client wrote to his lawyer describing the transaction in dispute, which the prosecution or the opposing party would dearly like to read. The letter is relevant on any view. It is also a confidential communication made in the course of and for the purpose of professional employment, and Sections 126 to 129 keep it out unless the client consents, with the two statutory exceptions, communications in furtherance of an illegal purpose and facts observed showing a crime or fraud committed since the employment began, marking the boundary. The fourth item therefore never reaches proof at all; it fails at admissibility for a reason that has nothing to do with reliability and everything to do with the law’s judgment that the institution of legal advice is worth more than the evidence the letter would have supplied. That is the deepest lesson the filter teaches, and the reason the three gates must be kept separate: the Act excludes material for at least three different kinds of reason, unreliability, informality and policy, and only a reader who can say which reason is operating in a given exclusion knows what argument might answer it.

Where the Act Does Not Bind: Tribunals, Inquiries, and Affidavits

Section 1 confines the Act to judicial proceedings in or before a court, and the territory outside that boundary is larger than most readers expect. Arbitrators are expressly outside it, which is why arbitral tribunals receive material under their own procedural discretion while often borrowing the Act’s distinctions for guidance. Affidavits presented to a court or officer are expressly outside it, since an affidavit is not testimony: its contents are regulated by the procedural codes and by the court’s power to order the deponent’s attendance for cross-examination, and a fact asserted on affidavit acquires the quality of evidence only when the process treats it so. Departmental and disciplinary inquiries occupy the most litigated part of the boundary. The Supreme Court held in State of Haryana v Rattan Singh in 1977 that the strict and sophisticated rules of evidence under the Evidence Act do not apply to departmental inquiries, which are governed by the principles of natural justice instead; the inquiry officer must act on material that is logically probative, a conclusion cannot rest on mere suspicion or on material the delinquent had no opportunity to meet, but hearsay is not automatically excluded and the enquiry is not a trial. The distinction that runs through this line of cases is between the rules of evidence and the principles behind them: a tribunal freed from the rules is not freed from the requirement that its findings rest on evidence of some probative value, and courts reviewing tribunal findings reach for Act-shaped concepts, relevance, burden, the difference between suspicion and proof, precisely because the Act’s grammar is the only fully articulated grammar of proof the system possesses. The filter therefore governs in two modes: as binding law inside the courtroom, and as the persuasive template by which every quasi-judicial process in the country is judged when its findings are challenged for want of evidence.

Commissions of inquiry appointed under the Commissions of Inquiry Act, 1952 illustrate the boundary from the other side. A commission is a fact-finding body and not a court; the Evidence Act does not govern its proceedings, and its report, however authoritative its findings may appear, is a recommendation to the government that appointed it rather than a judgment binding any person. The 1952 Act itself supplies the evidential protection that makes candour before a commission possible: Section 6 of that Act provides that no statement made by a person to a commission shall subject him to, or be used against him in, any civil or criminal proceeding except a prosecution for giving false evidence before the commission. Evidence gathered by a commission therefore does not migrate into a later trial through the commission’s file; a prosecution that follows must build its case again through the Act’s own gates, examining its witnesses in court in the ordinary way. The pattern completes the picture of the Act’s reach: inside the courtroom it is the law of proof; outside it, it is the standard by reference to which the products of other processes are refused entry until they have been re-proved in its own form.

The Amendment Arc and the Replacement of 2023

For a code that governed for a century and a half, the Evidence Act was amended sparingly, and the amendments cluster at the pressure points this article has identified. The first great cluster belongs to the criminal law reforms of the 1980s: Section 113A in 1983, attaching the abetment presumption to suicides of married women within seven years of marriage where cruelty is shown, and Section 113B in 1986, attaching the mandatory dowry death presumption, each enacted alongside penal provisions creating the corresponding offences, and each a legislative judgment that ordinary proof rules were failing in the closed domestic setting where these deaths occur. The second cluster is the electronic one: the Information Technology Act, 2000 rebuilt the Act’s perimeter for the digital record, inserting Sections 65A and 65B, the definitions that carried electronic records into documentary evidence, the opinion provisions for electronic signatures, and documentary presumptions for electronic agreements, messages and five-year-old electronic records in Sections 85A to 85C, 88A and 90A, with the Examiner of Electronic Evidence following in Section 45A under the 2008 amendment of the technology statute. The third cluster is the Criminal Law (Amendment) Act, 2013, which inserted Section 53A, making evidence of a victim’s character or previous sexual experience irrelevant on the issue of consent in specified sexual offence prosecutions, together with the Section 114A consent presumption and protective adjustments to the cross-examination provisions. Each cluster amended the filter without redesigning it, and the pattern is itself evidence about the original design: a legislature that needed a new gate built one onto the frame, and a legislature dissatisfied with the frame would have had to do what the Parliament of 2023 eventually chose to do.

The Bharatiya Sakshya Adhiniyam, 2023 replaced the Act as part of the three-code recasting of Indian criminal law, receiving assent as Act 47 of 2023 and entering into force on 1 July 2024, with proceedings already instituted carried forward under the repeal and savings provision of the new code. The successor article in this series owns the replacement in detail, and the treatment here is confined to the question this article’s own argument raises: what did the replacement do to the filter? The honest answer is that it preserved the architecture and modernised the fittings. The three-gate sequence survives intact, the relevancy classes were carried across with renumbering, and electronic records were promoted from a special regime bolted onto documentary evidence into the centre of the definition of a document, with the successor’s provisions treating material stored across multiple files as original in each of its instances. On the certificate question, the debate that ran from Anvar to Arjun Panditrao, the assumption that replacement would abolish the certificate turns out to be wrong: the successor’s Section 63 retains the requirement for secondary evidence of electronic records and adds a prescribed certificate in the Schedule to the new code, divided between a part completed by the party producing the record and a part completed by an expert or the person in charge of the device. The gate, in other words, was judged sound by the replacing legislature; what changed is the paperwork at the gate, and the litigation that follows will test whether the prescribed forms reduce the satellite disputes about defective certificates or multiply them.

Assessment: What the Filter Does Well, and Where It Strains

The strengths of the Act are the strengths of its design decision in 1872. A closed list of relevant facts, stated in provisions a district judiciary could actually apply, gave an enormous and uneven court system a common language for the most frequent rulings in litigation, and the language proved stable enough that a case decided on Section 65B’s ancestor provisions or on the confession sections a century ago still speaks directly to a hearing in the electronic age, through a chain of interpretation rather than a succession of codes. The confession rules, for all the argument about Section 27, reflect a legislative distrust of custodial statements that the subsequent record of custodial abuse in India has done nothing to discredit, and the presumption machinery remains a model of how a code can decide in advance, transparently and revisably, who bears the cost of missing proof. The Act also disciplines its own subject: by separating relevancy, admissibility, proof and appreciation, it gives appellate courts a vocabulary for locating error precisely, and gives advocates a sequence in which to build and attack a case.

The strains are equally identifiable, and they are not primarily the strains of age. The certificate litigation of the last decade was not caused by a statute too old to comprehend a compact disc; it was caused by a provision inserted in 2000 whose mandatory character two benches read differently, and the decade of satellite litigation that followed says more about the costs of interpretive inconsistency than about Victorian drafting. The genuine structural strains lie elsewhere. Section 27 continues to sit uneasily beside Sections 25 and 26, and its administration depends on a vigilance about staged discoveries that the reported record shows to be uneven. The Act’s appreciation stage, where outcomes are actually decided, is necessarily the least codified, and no drafting can fully protect a system in which the same sole witness is believed in one courtroom and disbelieved in another; that is a feature of adjudication the code manages rather than solves. And the filter’s documentary heartland, built for a world of paper originals and copies, continues to absorb material, messaging threads, platform records, machine-generated logs, whose properties the statutory concepts of original and copy fit awkwardly even after the successor code’s repairs.

The verdict follows from the analysis rather than from affection for an old statute. The Evidence Act succeeded at the task Stephen set it: it made the rulings that decide what a court may hear predictable enough to be argued about in a shared vocabulary, and durable enough that the argument could accumulate for a century and a half without the frame giving way. Its failures are located, specific and instructive: a discovery exception that must be policed to remain honest, a certificate gate whose cost is measured in satellite litigation, and an appreciation stage that no code can save from the quality of the judges who operate it. The reader who wants the working detail of the successor regime should turn to the treatment of the Bharatiya Sakshya Adhiniyam in this series, and the reader who wants the confession and procedure machinery in operation should turn to the criminal procedure pillar; this article’s claim is the one it began with, that between every fact and every finding in an Indian courtroom stand three separate gates, and that knowing which gate a piece of evidence failed at is the beginning of knowing what to do next.

Frequently Asked Questions

Q: Who drafted the Indian Evidence Act and when?

The Indian Evidence Act was drafted by James Fitzjames Stephen, who served as Law Member of the Viceroy’s Council from 1869 to 1872. Stephen, who distrusted the sprawling case-law form of English evidence law, reduced the subject to a compact code organised around the relevancy of facts. His draft was introduced in 1871, passed as Act 1 of 1872, and brought into force on 1 September 1872. An earlier draft associated with his predecessor, Henry Maine, had been set aside as unsuited to Indian conditions. Stephen later distilled the same scheme into his English Digest of the Law of Evidence, but the Indian statute came first and governed Indian courts for a century and a half.

Q: What is the difference between relevancy and admissibility under the Indian Evidence Act?

Relevancy is the first question the Act asks: whether a fact belongs to one of the classes Sections 5 to 55 declare a court may reason from, or is a fact in issue itself. Admissibility is a second, narrower question asked of facts that are relevant: whether some exclusionary rule, privilege or formal requirement keeps the particular item out despite its relevance. A confession to a police officer is the standard illustration, plainly relevant to guilt and flatly inadmissible under Section 25. The two questions are decided at different points and for different reasons, and they are followed by a third, proof, which asks whether the admitted material has actually been established in the manner the Act requires.

Q: Why does the Indian Evidence Act bar a confession made to a police officer?

Section 25 provides that no confession made to a police officer shall be proved against a person accused of an offence, and Section 26 extends the bar to confessions made by a person in police custody to anyone, unless made in the immediate presence of a Magistrate. The reason is institutional distrust hardened into rule. A confession obtained by custodial pressure is easy to extract and difficult to test, because the trial contest would collapse into oath against oath between the accused and the officer. The legislature of 1872 chose to sacrifice even reliable confessions to the police rather than litigate coercion in every case, and directed confessions instead into the Magistrate-recorded channel, where warning and reflection are built into the process.

Q: What is a dying declaration and why is it accepted?

A dying declaration is a statement made by a person as to the cause of his death, or as to any of the circumstances of the transaction which resulted in his death, in a case where the cause of that person’s death comes into question. Section 32(1) makes such a statement relevant even though the maker is dead and can never be cross-examined, and it adds that the statement is relevant whether or not the maker was under expectation of death when making it. The statement is accepted because the alternative would silence the victim of a fatal attack in precisely the cases where other witnesses are least likely to exist. Acceptance is not credulity: the courts test the declaration for truth, voluntariness, fitness of mind and consistency, and a declaration that survives that scrutiny can by itself sustain a conviction.

Q: How did the Indian Evidence Act treat electronic records?

The original Act of 1872 knew nothing of electronic records, and the treatment arrived with the Information Technology Act, 2000. That amendment brought electronic records within the definition of documentary evidence in Section 3, barred oral admissions as to the contents of electronic records unless genuineness is in question under Section 22A, and inserted Sections 65A and 65B as a special code for proving them. Section 65B treats computer output as a document admissible without production of the original if conditions about the regular use and proper operation of the computer are satisfied and a certificate under Section 65B(4) is produced. The amendment also added opinion and presumption provisions for electronic material, including the opinion of the Examiner of Electronic Evidence in Section 45A.

Q: What did Anvar and Arjun Panditrao decide about certificates?

In Anvar P V v P K Basheer in 2014, a three-judge bench held that Section 65B is a complete code for electronic evidence, so that an electronic record relied on as secondary evidence is admissible only with the certificate Section 65B(4) requires, and the earlier permissive approach in the Navjot Sandhu case was not good law. A two-judge bench in Shafhi Mohammad in 2018 relaxed the requirement for parties not in possession of the device. In Arjun Panditrao Khotkar v Kailash Kushanrao Gorantyal in 2020, a three-judge bench on a reference restored Anvar, overruled Shafhi Mohammad, and held the certificate to be a condition precedent, while clarifying that it may be produced at a later stage of the trial and that a party locked out of the device may ask the court to summon the person in charge of it.

Q: What is the presumption in a dowry death case?

Section 113B provides that where the question is whether a person has caused the dowry death of a woman, and it is shown that soon before her death the woman was subjected by that person to cruelty or harassment for, or in connection with, a demand for dowry, the court shall presume that the person caused the dowry death. The presumption is mandatory once the foundation facts are proved: the death within the scope of the offence, the cruelty or harassment, its connection with a dowry demand, and its proximity in time to the death, expressed in the phrase soon before. The prosecution must prove every element of the foundation; the presumption then supplies the causal element, and the accused may displace it by showing the contrary on the evidence.

Q: What does declaring a hostile witness actually mean?

The Act never uses the expression hostile witness. The situation arises under Section 154, which allows the court, in its discretion, to permit the party who called a witness to put to that witness any questions that could be put in cross-examination by the adverse party, including leading questions and questions about previous statements. A witness is treated this way when he resiles from his earlier account or appears unwilling to tell the truth for the side that called him. The permission does not brand the witness a liar or erase his testimony. In Sat Paul v Delhi Administration in 1976, the Supreme Court held that the evidence of such a witness remains evidence and may be relied on by either side to the extent the court finds it trustworthy after the testing the permission allows.

Q: Who bears the burden of proof under the Indian Evidence Act, and how is it allocated?

Section 101 places the burden on the person who asks the court to give judgment on a legal right or liability dependent on facts he asserts: he must prove those facts exist. Section 102 locates the burden on the party who would fail if no evidence at all were given on either side, and Section 103 puts the burden of any particular fact on the person who wishes the court to believe in it. In a criminal trial this means the prosecution bears the burden on the offence throughout and it never shifts, while the burden of individual facts moves as evidence accumulates. Section 105 makes a defined exception: an accused who claims the benefit of a general or special exception bears the burden of proving the exculpating circumstances, measured by the civil standard of preponderance of probabilities.

Q: What does Section 106 mean by a fact especially within a person’s knowledge?

Section 106 provides that when any fact is especially within the knowledge of any person, the burden of proving that fact is upon him. The classic illustration is the passenger charged with travelling without a ticket, who alone knows whether he paid the fare. The section does not relieve the prosecution of its general burden. In Shambhu Nath Mehra v State of Ajmer in 1956, the Supreme Court held that Section 106 is an exception to the general rule in Section 101 and applies only where the fact is impossible or disproportionately difficult for the prosecution to prove and comparatively simple for the accused to explain. The prosecution must first prove the setting that makes the fact specially the accused’s to explain; only then does the section place the burden of that fact on him.

Q: What is the difference between an admission and a confession under the Indian Evidence Act?

An admission, defined by Section 17, is any statement by a party or a person whose position is in question that suggests an inference as to a fact in issue or a relevant fact. A confession is a species of admission narrowed to criminal proceedings, and although the Act does not define it, the Privy Council in Pakala Narayana Swami v Emperor in 1939 held that a confession must either admit the offence in terms or admit substantially all the facts constituting the offence, and that a statement containing self-exculpatory matter which, if true, would negate the offence is not a confession. The distinction carries consequences: admissions are governed by the comparatively mild rules of Sections 17 to 23, while confessions attract the exclusionary screens of Sections 24 to 27, including the bar on confessions made to police officers.

Q: What is estoppel under Section 115 of the Indian Evidence Act?

Section 115 provides that when one person has, by declaration, act or omission, intentionally caused or permitted another person to believe a thing to be true and to act upon that belief, neither the first person nor his representative in interest shall be allowed, in any suit or proceeding between them, to deny the truth of that thing. The stock illustration is the tenant, who is not permitted during the tenancy to deny that his landlord had title when the tenancy began. Estoppel under the Act is a rule of evidence rather than a cause of action: it does not create rights, it seals a party’s mouth about a fact on which his own conduct induced the other party to rely, and it operates only between the parties to that reliance.

Q: When can secondary evidence of a document be given under Section 65?

Section 64 requires a document to be proved by primary evidence, the document itself, and Section 65 permits secondary evidence of its contents on seven grounds: the original is in the possession of the adversary or of a person out of reach or not subject to the court’s process; the adversary has admitted the original’s existence and contents in writing; the original has been destroyed or lost, or cannot be produced in reasonable time for a reason not arising from the party’s own default or neglect; the original is not easily movable; the original is a public document; the original is a document of which a certified copy is permitted to be given in evidence; or the case concerns numerous accounts that cannot conveniently be examined in court. Secondary evidence of a document in the adversary’s hands ordinarily requires prior notice to produce under Section 66.

Q: What is the difference between may presume, shall presume and conclusive proof?

Section 4 supplies three registers of presumption. Where the Act says the court may presume a fact, the court may either regard the fact as proved unless and until it is disproved, or call for proof of it; most presumptions in Section 114 and the document presumptions of Part II sit in this discretionary register. Where the Act says the court shall presume a fact, the court must regard it as proved unless and until it is disproved, as with the dowry death presumption in Section 113B. Where the Act declares one fact to be conclusive proof of another, the court must, on proof of the first, regard the second as proved and must not allow evidence to disprove it, as with legitimacy under Section 112. The three registers decide who loses when proof runs out, and with what degree of compulsion.

Q: How much weight does a court give to expert opinion under Section 45?

Section 45 makes an expert’s opinion a relevant fact on points of foreign law, science or art, and the identity of handwriting or finger impressions, but relevance is not weight. The opinion is advisory in character: it does not bind the court, it is not proof of the facts on which it rests, and its persuasive force depends on the expert’s qualifications, the material examined, the reasoning the expert exposes, and how the opinion survives cross-examination aimed at those foundations. A bare conclusion signed by a designated expert carries little; a reasoned opinion that shows its working carries more. The court may also compare writings itself under Section 73, and the cases counsel particular caution before resting a finding on handwriting comparison alone, since that discipline is among the least exact the section admits.

Q: Why is a confession obtained by inducement, threat or promise irrelevant under Section 24?

Section 24 makes a confession irrelevant where it appears to the court to have been caused by an inducement, threat or promise having reference to the charge, proceeding from a person in authority, and sufficient to give the accused grounds to suppose he would gain an advantage or avoid an evil of a temporal nature in the proceedings. The ground of exclusion is reliability through a specific channel: a person bargaining with authority for advantage may confess to what did not happen, so the law refuses the product of the bargain. The bar is carefully bounded. Section 28 restores a confession made after the impression of the inducement has been fully removed, and Section 29 preserves confessions otherwise relevant merely because they were made under a promise of secrecy, through deception, in drunkenness, or without a warning.

Q: What presumption applies to a document that is thirty years old?

Section 90 provides that where a document purporting or proved to be thirty years old is produced from custody which the court, in the particular case, considers proper, the court may presume that the signature and every other part of the document purporting to be in the handwriting of any particular person is in that person’s handwriting, and, in the case of a document executed or attested, that it was duly executed and attested by the persons by whom it purports to be executed and attested. The presumption is discretionary and attaches to the document’s authenticity of execution, not to the truth of its contents, which must still be proved. Its function is practical: after thirty years the attesting witnesses are commonly dead or untraceable, and without the presumption most old title documents would be unprovable.

Q: What is a leading question, and when may one be asked?

Section 141 defines a leading question as a question suggesting the answer which the person putting it wishes or expects to receive. Section 142 provides that leading questions must not, if objected to by the adverse party, be asked in examination-in-chief or in re-examination without the leave of the court, though the court permits them on matters that are introductory, undisputed, or already sufficiently proved. Section 143 permits leading questions in cross-examination without restriction of that kind. The scheme follows the structure of the trial: a party’s own witness is assumed to be favourable and must not be fed answers, while cross-examination is the testing stage, where the questioner is entitled to put the answer to the witness and require him to accept or reject it.

Q: What powers does a judge have to ask questions under Section 165?

Section 165 empowers the judge, in order to discover or obtain proper proof of relevant facts, to ask any question he pleases, in any form, at any time, of any witness or of the parties, about any fact relevant or irrelevant, and to order the production of any document or thing. Neither party may object to such a question or order, and neither may cross-examine a witness on an answer given to the judge without the court’s leave. The power is wide but not unbounded: the judgment must rest on facts that are relevant and duly proved, and the judge may not compel a witness to answer a question the witness could lawfully refuse, or dispense with primary evidence of a document where primary evidence is required. The section makes the Indian trial judge an active examiner rather than a passive referee.

Q: What is the difference between direct evidence and circumstantial evidence?

Direct evidence is evidence that, if believed, establishes the fact in issue without any intervening inference: the eyewitness who saw the accused strike the blow. Circumstantial evidence establishes other facts from which the fact in issue is inferred: the accused’s fingerprint on the weapon, his flight, his possession of the proceeds. The Act does not use either label as a term of art; its machinery of relevant facts in Sections 6 to 11 is what admits the links of a circumstantial chain. Because an inference does the work perception does in direct evidence, the courts impose a discipline stated in the five principles of Sharad Birdhichand Sarda v State of Maharashtra in 1984: the circumstances must be fully established, consistent only with guilt, conclusive in tendency, excluding every other hypothesis, and forming a chain so complete that no reasonable ground for innocence remains.