Tendering Pardon to an Accomplice in India: A Critical Analysis of Approver Evidence, Procedural Safeguards and Fair Trial under the BNSS, 2023

Introduction

A. Background of the study

Criminal offences are frequently committed through groups, conspiracies and organised networks rather than by isolated individuals. Where the agreement is secret, external witnesses may see only one act, and documents may not reveal who planned the offence, issued instructions, divided responsibilities or concealed the proceeds. The complete narrative may therefore be known only to a participant.

The law of tendering pardon addresses this evidentiary difficulty. It permits the prosecution to obtain testimony from a person directly or indirectly concerned with an offence. The person receives a conditional pardon and must make a full and true disclosure of the circumstances within his knowledge relating to the offence and to every other person involved. After acceptance, he becomes an approver and gives evidence for the prosecution.¹

The device is useful in conspiracy, organised crime, corruption, smuggling and intricate financial offences. An insider may reveal the structure of a criminal organisation, the agreement between participants, the allocation of tasks and the route by which money or property moved. Such evidence can convert disconnected circumstances into a coherent account.

The mechanism simultaneously creates a serious danger. The approver is not an impartial observer. He has participated in the offence and receives a substantial benefit, normally protection from prosecution or a more favourable procedural position. He may minimise his own conduct, exaggerate another person’s role or repeat a version developed during investigation. The law must consequently balance the public interest in prosecuting serious offenders with the accused’s right to a fair trial based on reliable and independently tested evidence.

The Bharatiya Nagarik Suraksha Sanhita, 2023, makes the subject especially important. Sections 343 to 345 substantially continue Sections 306 to 308 of the Code of Criminal Procedure, 1973.² The numbering has changed, but the earlier jurisprudence remains relevant because the essential statutory language and purpose are substantially similar. Courts must now apply that jurisprudence together with the Bharatiya Sakshya Adhiniyam, 2023, whose provision on accomplice evidence is not, as will be seen, a verbatim reproduction of the provision it replaces.

Judicial pardon must also be distinguished from executive clemency under Articles 72 and 161 of the Constitution. Executive clemency ordinarily follows conviction and concerns remission, commutation or reduction of sentence. Tender of pardon under the BNSS is a procedural and evidentiary device operating during investigation, inquiry or trial. Its purpose is to secure evidence against other participants.

B. Statement of the research problem

The central problem is whether the BNSS adequately protects the accused while permitting the State to exchange conditional immunity for accomplice evidence. The statute requires full and true disclosure, but it does not provide a detailed protocol for determining voluntariness, ensuring legal assistance, recording informal assurances, reviewing custody or identifying the minimum corroboration needed for conviction.

The problem is acute when the proposed approver is in custody. A person facing a serious charge may accept pardon because of fear of conviction, prolonged detention or pressure from investigators. A written acceptance does not itself establish that the decision was free and informed. The court must distinguish a genuine decision to disclose the truth from a statement produced by custodial vulnerability.

A second problem concerns the relationship between prosecution and court. The prosecution commonly initiates a pardon request because it knows the investigative value of the accomplice. Yet the court must exercise independent judicial discretion. If the prosecution’s request becomes decisive, the judge may be reduced to a formal approving authority.

A third problem is corroboration. Section 138 of the BSA treats an accomplice as a competent witness and states that a conviction is not illegal if it proceeds upon the corroborated testimony of an accomplice. Section 119, Illustration (b), permits the court to presume that an accomplice is unworthy of credit unless he is corroborated in material particulars.³ The practical meaning of “material” and “connecting” corroboration remains principally judicial.

C. Research questions and objectives

This study asks whether Sections 343 to 345 adequately regulate tender of pardon; whether approver testimony should support conviction without independent corroboration; whether existing safeguards protect against coercion, prosecutorial manipulation and undisclosed bargains; how CrPC precedents should operate under the BNSS; and whether legislative or judicial reform is necessary.

Its objectives are to examine the statutory scheme, analyse approver evidence under the BSA, evaluate Supreme Court safeguards, identify risks of self-serving testimony and custodial pressure, and distinguish legitimate use in secret offences from improper use where independent evidence is already sufficient. The study also proposes reforms reconciling investigative necessity with personal liberty and fair trial.

D. Methodology, scope and limitations

The research adopts a doctrinal and analytical methodology. It examines the BNSS, the BSA, the Constitution and judicial decisions, while using secondary literature to understand the history and policy of accomplice evidence. It does not statistically analyse every pardon application in India. Its focus is the legal structure and judicial control of tendering pardon.

The study concerns judicial pardon under Sections 343 to 345 of the BNSS. Executive powers under Articles 72 and 161 are mentioned only for distinction. The corresponding CrPC provisions are examined because leading Indian decisions were delivered under them and remain persuasive where the BNSS retains the same legal design.

Conceptual and historical foundations

A. Accomplice and approver

An accomplice is a person who participates in an offence or is knowingly connected with it. Participation may involve planning, financing, facilitating, assisting, executing or concealing the crime. The precise form of participation depends on the substantive offence. An accomplice is not disqualified from testifying merely because of that connection, but participation creates a risk of bias and self-interest.⁴

An approver is an accomplice who accepts a tender of pardon on the condition of making a full and true disclosure. Acceptance changes his procedural position from accused or suspect to prosecution witness, but it does not erase his former participation. The pardon is conditional, not an unconditional declaration of innocence. If he deliberately conceals material facts or gives false evidence, he may be tried for the offence in respect of which the pardon was tendered.⁵

The distinction between status and credibility is fundamental. An approver becomes competent to testify, but the grant of pardon does not make his account inherently trustworthy. The court must examine his motive, opportunity for knowledge, consistency, benefit received, conduct and conformity with objective evidence.

B. Historical development

The common-law system permitted accomplice testimony because secret offences could otherwise remain unproved. Courts nevertheless distrusted a criminal who accused another person to save himself. The response was to admit the evidence while developing a rule of prudence requiring independent corroboration.

In R. v. Baskerville, the English Court of Criminal Appeal explained that corroboration must be independent testimony connecting, or tending to connect, the accused with the commission of the offence.⁶ Indian law adopted the same broad approach. The former Indian Evidence Act treated an accomplice as competent, while courts ordinarily required corroboration; the CrPC provided a judicial mechanism for tendering pardon. The BNSS and the BSA continue that policy, although the BSA states the evidentiary rule in altered language, as Chapter IV explains.

The historical compromise is therefore clear: the State obtains evidence by conferring a conditional advantage on one participant, but reliability depends on transparency, limited discretion and judicial scrutiny. The mechanism is justified by truth-seeking necessity, not by a general policy that one offender should be rewarded for accusing another.

C. Public interest and evidentiary necessity

The strongest justification for pardon is that serious offences may be impossible to prove without an insider. In organised crime, the approver may identify the hierarchy, instructions, financing and distribution of work. In corruption, he may explain the relationship between officials, intermediaries and beneficiaries. In economic offences, he may reveal manipulated accounts, hidden transactions and transfers of proceeds.

The public interest, however, is not identical with obtaining a conviction at any cost. Pardon is appropriate only when the proposed testimony is material and genuinely needed. If the prosecution already possesses reliable independent evidence, granting immunity may merely allow one participant to escape punishment. The court should consider the evidentiary value of the proposed disclosure, the person’s knowledge and whether his account can be independently verified.

In Suresh Chandra Bahri v. State of Bihar, the Supreme Court described the dominant object of the provision as preventing offenders in grave cases from escaping punishment where no clue or trace of the offence is otherwise available.⁷ That object must be pursued consistently with fair procedure and evidentiary caution.

D. Inherent risks

An approver may minimise his own responsibility by portraying himself as a reluctant or minor participant. He may falsely implicate a co-accused because of hostility, fear or an expectation of favourable treatment. He may selectively disclose facts that support the prosecution while concealing the involvement of influential associates. Prolonged custody may produce indirect pressure even without an express threat.

These risks do not justify excluding all accomplice evidence. They justify a demanding process. The court must insist on a candid account of the approver’s own role, test the chronology against objective records and identify independent evidence linking the accused with the offence. The more substantial the benefit to the approver, the more careful the scrutiny should be.

Statutory framework under the BNSS, 2023

A. Section 343: authority and conditions

Section 343 authorises the Chief Judicial Magistrate, at any stage of the investigation or inquiry into or the trial of an offence, and a Magistrate of the first class inquiring into or trying the offence, to tender a pardon to a person supposed to have been directly or indirectly concerned in, or privy to, an offence, with a view to obtaining his evidence.⁸ The essential condition is that the person must make a full and true disclosure of the whole of the circumstances within his knowledge relating to the offence and to every other person concerned, whether as principal or abettor.

“Full” requires disclosure of all material facts within the person’s knowledge, including his own conduct and the participation of every other person. “True” requires honesty and prohibits deliberate distortion. A selected account that gives the prosecution useful facts while concealing material circumstances does not satisfy the condition.

The provision applies to offences triable exclusively by the Court of Session or by the Court of a Special Judge, and to offences punishable with imprisonment which may extend to seven years or with a more severe sentence.⁹ The limitation reflects the exceptional nature of exchanging immunity for evidence. Seriousness is necessary but not sufficient; the court must still assess necessity, voluntariness and suitability.

B. Competent authorities and judicial discretion

The statutory distribution of power between the Chief Judicial Magistrate and the Magistrate of the first class prevents the investigating agency from creating an informal immunity arrangement. One drafting change should be noted. Section 306(1) of the CrPC also named the Metropolitan Magistrate; the BNSS has abolished that class of court altogether, and Section 343(1) accordingly names only the Chief Judicial Magistrate and the Magistrate of the first class.¹⁰ The pardon must be tendered by the legally designated authority at the appropriate procedural stage.

The power is judicial, not executive. Investigators and prosecutors may request or support pardon, but they cannot grant it. The court must apply its own mind to the seriousness of the offence, the materiality of the expected testimony, the proposed approver’s knowledge and the fairness of the process.

The Magistrate must record his reasons for tendering pardon and whether the tender was accepted, and must furnish the accused with a copy of that record free of cost on application.¹¹ A reasoned order should identify the offence, the apparent role of the proposed approver, the facts expected from him and why the evidence is necessary. A bare reference to the interest of justice is inadequate. Reasons enable review, protect the accused from arbitrary selection and inform the approver about the obligation undertaken.

In Lt. Commander Pascal Fernandes v. State of Maharashtra, the Supreme Court stressed that the judge must know the nature of the evidence the person is likely to give, the nature of his complicity and the degree of his culpability before tendering pardon, so that the power is exercised for the statutory purpose of obtaining evidence and not as an unstructured act of favour.¹²

C. Acceptance and examination

Acceptance should be voluntary and informed. The Magistrate should explain, in a language understood by the proposed approver, that the pardon is conditional; that he must disclose the entire truth; that he will become a prosecution witness; that he may be cross-examined; and that breach may result in prosecution. The court should record that the person had a genuine choice to refuse.

A written signature is not conclusive proof of voluntariness. The court should examine the circumstances of custody, the availability of legal assistance, the timing of the request and whether any additional promise was made. The person must understand the distinction between an investigative statement and evidence given after accepting pardon.

The approver must be examined as a witness before the Magistrate taking cognizance and in the subsequent trial. This requirement places the account on the judicial record and permits the accused to test it. The earlier record also enables comparison with later testimony, exposing material omissions and improvements. The investigating agency cannot substitute an informal confession or police statement for the statutory examination.

State of Maharashtra v. Abu Salem Abdul Kayyum Ansari confirms how seriously that obligation is taken. A person who accepts pardon ceases to be an accused and becomes a witness for the prosecution, and so long as the Public Prosecutor has not certified non-compliance, the prosecution is bound to examine him as a witness both in the committing court and in the trial court.¹³ The requirement is substantive because it protects both evidentiary reliability and the defence’s opportunity to challenge the witness.

D. Custody

The statute directs that a person who accepts a tender of pardon shall, unless he is already on bail, be detained in custody until the termination of the trial.¹⁴ Custody can prevent absconding, witness influence and interference with justice, but it must not become a means of forcing acceptance or maintaining a favoured version.

The court should periodically review custody, consider changed circumstances and ensure that investigators do not exercise continuing influence after pardon. Article 21 requires that deprivation of liberty follow a procedure that is fair, just and reasonable.¹⁵ Continued detention must serve a legitimate purpose and cannot operate as an indirect compulsion to testify.

E. Section 344: court-directed pardon

Section 344 operates after the case has left the Magistrate. At any time after commitment of a case but before judgment is passed, the court to which the commitment is made may itself tender a pardon on the same condition to a person supposed to have been directly or indirectly concerned in, or privy to, the offence, with a view to obtaining his evidence at the trial.¹⁶ The marginal heading speaks of a power to direct the tender of pardon, but the power conferred is a power in the committal court to make the tender. An accused may request pardon, but the request creates no automatic right. The prosecution should ordinarily be heard because it understands the investigative position, yet its consent cannot bind the court.

The court must determine whether the proposed evidence is material, whether the person can make a complete disclosure and whether pardon is genuinely necessary. It should not grant pardon merely to reduce the number of accused or because one participant seeks a personal benefit.

In C.B.I. v. Ashok Kumar Aggarwal, the Supreme Court rejected the proposition that pardon must invariably be granted whenever the prosecution supports it.¹⁷ Such a rule would transform judicial discretion into prosecutorial entitlement.

F. Section 345: breach

Section 345 reflects the conditional character of pardon. The provision is set in motion by a certificate of the Public Prosecutor that the person has, either by wilfully concealing anything essential or by giving false evidence, not complied with the condition on which the tender was made. He may then be tried for the offence in respect of which the pardon was tendered, for any other offence of which he appears to have been guilty in the same matter, and for giving false evidence, though he may not be tried jointly with any other accused and a trial for giving false evidence requires the sanction of the High Court.¹⁸ The burden does not lie on him: he is entitled to plead that he complied with the condition, in which case it is for the prosecution to prove that the condition was not complied with, and the court must find on that question before passing judgment and must acquit if it finds compliance.¹⁹

Minor inconsistency, imperfect memory or an immaterial omission should not automatically constitute breach. The approver must receive notice and an opportunity to explain. A reasoned decision must distinguish intentional deception from ordinary human fallibility.

Evidentiary value of approver testimony

A. Competence under the BSA

Section 138 of the BSA declares an accomplice to be a competent witness against an accused person and states that a conviction is not illegal if it proceeds upon the corroborated testimony of an accomplice.²⁰ The provision recognises practical necessity: without insider evidence, organised or secret offences may be impossible to establish. It does not, however, equate competence with credibility. The wording also repays attention, because it is not that of the provision it replaces. Section 133 of the Indian Evidence Act, 1872, provided that a conviction was not illegal merely because it proceeded upon the uncorroborated testimony of an accomplice; Section 138 of the BSA drops “merely” and speaks instead of corroborated testimony.²¹ On its face the new text states the sufficiency of corroborated accomplice evidence rather than the legality of an uncorroborated conviction, and the point has not yet been settled by authority.

Section 119, Illustration (b), permits the court to presume that an accomplice is unworthy of credit unless he is corroborated in material particulars.²² Read together, the provisions create a balance. The witness may be heard, but the court must recognise the risk that a person seeking immunity may falsely implicate others.

The distinction between legal sufficiency and judicial prudence is important. A conviction is not automatically illegal merely because the witness is an accomplice. Nevertheless, courts ordinarily require corroboration because the witness has a powerful motive for self-protection. The prudent approach is not a rigid demand that every sentence be independently confirmed; it is a requirement that independent evidence connect the accused with the important features of the offence.

B. Corroboration

Corroboration must be independent and material. Evidence merely proving that a crime occurred is insufficient if it does not connect the accused with participation. Supporting material may include documents, electronic records, bank transactions, call data, recoveries, admissions, conduct or testimony from an independent witness.

If the approver states that the accused attended a meeting and issued instructions, corroboration may include communications, location records, documents prepared during the meeting or subsequent conduct consistent with the agreement. If he alleges delivery of illegal money, bank records, recovery or a contemporaneous record may provide support.

The trial court should identify the exact corroborative evidence and explain how it supports the relevant part of the testimony. A general statement that the witness is corroborated is not sufficient. Corroboration need not independently prove every detail, but it must reduce the possibility of false implication.

In Sarwan Singh Rattan Singh v. State of Punjab, the Supreme Court applied a double test to approver evidence: the court must first satisfy itself that the approver is a reliable witness, and only then look for corroboration connecting the accused with the crime in material particulars.²³ The order of the inquiry matters, because corroboration cannot repair an account the court has no reason to believe in the first place. The principle remains applicable under Section 138 of the BSA.

The approver’s earlier statement ordinarily cannot corroborate his trial evidence because both originate from the same interested source. Corroboration should arise from evidence independent of the approver. The court should also avoid treating investigative consistency alone as sufficient where the investigation itself may have shaped the account.

C. Credibility and motive

The court should examine the benefit received or expected, the witness’s own admission of participation, his motive to implicate the accused, the circumstances in which the statement was made, and the consistency of his chronology. Candid acknowledgment of personal involvement may support credibility, but it does not eliminate the need for independent support.

Timing is relevant. A disclosure made soon after the person decided voluntarily to cooperate may stand differently from a version produced after repeated interrogation, threats or negotiations. The court should compare the first account with later statements and trial testimony. A material improvement that introduces a new role or incriminating circumstance may indicate reconstruction.

The witness must speak from personal knowledge. An approver cannot convert hearsay, assumption or information received from another person into direct evidence merely by presenting it confidently. The court should separate observed facts from inference and speculation.

D. Cross-examination and disclosure

Cross-examination is the principal safeguard against unreliable approver evidence. The defence should be allowed to ask about the witness’s own involvement, the benefit received, custody, prior statements, contradictions, omissions, hostility and reasons for implicating the co-accused. The accused should also challenge whether the witness had personal knowledge of the alleged events.

Effective cross-examination requires disclosure of the pardon order, relevant statements, reasons for granting pardon and every material promise or assurance connected with the arrangement. A hidden bargain prevents the defence from exposing motive and makes judicial evaluation incomplete. The right to a fair trial includes a meaningful opportunity to test prosecution evidence.²⁴

E. Circumstantial cases and conspiracy

Many pardon cases depend substantially on circumstantial evidence. The approver may supply a link in the chain, but his assertion cannot fill every gap. Independent evidence should support the important links, and the prosecution must establish a complete chain consistent with guilt.

In conspiracy cases, the evidence must establish knowledge of the unlawful purpose and participation in furtherance of it. Association, acquaintance or presence at a location does not automatically prove agreement. The approver should identify specific acts, communications or instructions demonstrating participation. Objective records are particularly valuable because they test the broad narrative often offered in secret-conspiracy cases.

F. Confession distinguished from approver evidence

An approver’s trial testimony is not identical to a confession. A confession is an admission of an offence; approver evidence is testimony given by a participant who has accepted conditional pardon and is examined as a prosecution witness. The safeguards applicable to confessions and those governing accomplice testimony should not be conflated.

The prosecution must comply with the applicable law before relying upon any earlier statement. The approver’s substantive account must be tested on oath through examination and cross-examination. An investigative statement may have a legally permitted use for contradiction or another limited purpose, but it cannot automatically replace evidence recorded in court.

Judicial approach and constitutional safeguards

A. Judicial independence

The power to tender pardon is discretionary. The prosecution may request it and an accused may seek it, but the court must decide whether the evidence is necessary and whether the conditions are fulfilled. The judge should not become an investigator searching for a witness to strengthen the prosecution, nor should the judge mechanically approve the investigative request.

In C.B.I. v. Ashok Kumar Aggarwal, the Supreme Court held that prosecutorial support does not compel the court to grant pardon; to hold otherwise would marginalise the role of the court and remove the judge’s discretion to ensure a fair trial. Restating the approach of Pascal Fernandes, the Court also required attention to the nature of the evidence expected, the proposed approver’s complicity and the degree of his culpability in relation to the co-accused, and it dismissed the appeal against the High Court order which had set the pardon aside and remitted the application for fresh consideration.²⁵ The decision protects the judicial role and recognises that a pardon order affects the accused, the proposed approver and the integrity of the trial.

B. Pascal Fernandes and the approver’s request

In Lt. Commander Pascal Fernandes v. State of Maharashtra, the Supreme Court considered whether an accused could himself seek pardon. It recognised that such a request may be entertained, while holding that it is ordinarily for the prosecution to ask that a particular accused be tendered pardon, and that where the accused applies directly the court should refer the request to the prosecution before acting on it.²⁶ The decision balances access to the mechanism with institutional discipline.

An accused therefore has no absolute right to pardon, but neither is he automatically barred from requesting it. The prosecution’s response assists the court, yet the final decision remains judicial. The process must not become a private bargain between the accused and the judge or a prosecutorial arrangement outside the record.

C. Procedural compliance

State of Maharashtra v. Abu Salem Abdul Kayyum Ansari shows how far the consequences of default reach. Where the Public Prosecutor certifies that the approver has not complied with the condition of the pardon, the pardon is forfeited, he reverts to the position of an accused, and his evidence has to be ignored in its entirety and does not remain legal evidence in the trial of the co-accused.²⁷ State through C.B.I., Chennai v. V. Arul Kumar maps the two routes by which a pardon may reach a Special Judge’s court. Where the charge-sheet is filed before a Magistrate, the Magistrate may tender pardon under Section 306 of the CrPC, even for an offence triable exclusively by a Special Judge, and then commit the case; where the Special Judge takes cognizance directly, Section 306 is bypassed and Section 307 applies, so that only the Special Judge may tender the pardon.²⁸

These decisions remain relevant under the BNSS, where the same division now falls between Sections 343 and 344. The change in statutory numbering does not authorise agencies or courts to disregard the procedural sequence. A correct record protects the defence’s right to know the evidence and enables the trial court to assess contradictions, omissions and improvements.

D. Articles 14, 20(3) and 21

Article 21 requires a fair, just and reasonable procedure.²⁹ Tendering pardon affects the approver’s liberty and the co-accused’s right to challenge evidence. The proposed approver must understand the condition and make a genuine choice; the co-accused must receive sufficient information for effective cross-examination.

Article 14 requires non-arbitrary public power.³⁰ Selection of one accomplice for pardon and another for prosecution should rest on evidentiary necessity, not favouritism, political influence or convenience. Relative culpability and evidentiary value are relevant. Where a central offender receives immunity and a less culpable person is prosecuted on his testimony, the court should demand particularly strong corroboration.

Article 20(3) protects an accused against compelled self-incrimination.³¹ Tender of pardon can operate consistently with that protection only when acceptance is voluntary. The Magistrate should explain the option to refuse, the duty of full disclosure and the risk of prosecution for breach. Prolonged custody, threats or improper promises must not convert the process into compelled testimony.

E. Corruption, economic offences and organised crime

In corruption and economic cases, an approver may reveal relationships among officials, intermediaries, contractors and beneficiaries, as well as the method of payment and concealment. These cases also generate substantial documentary and electronic evidence. Bank statements, digital communications, call records, accounting material and recoveries should therefore be used to test the account rather than treating the approver as a substitute for investigation.

In organised crime and conspiracy, an insider may identify the hierarchy and agreement, but the prosecution must prove knowing participation. Mere communication or association is insufficient. The court should require objective evidence supporting important allegations, especially where the approver’s narrative is the principal link between the accused and the offence.

Critical analysis, findings and recommendations

A. Adequacy of the framework

The BNSS framework is valuable because it preserves a method for exposing crimes committed in secrecy. The requirements of full and true disclosure, recorded reasons, examination and consequences of breach are significant protections. Nevertheless, the framework is incomplete. It does not prescribe a formal voluntariness inquiry, mandatory recording of all benefits and assurances, uniform custody review, legal assistance or a detailed corroboration standard.

This absence may produce inconsistent protection. One court may investigate the circumstances carefully, while another may treat pardon as a routine administrative step. Reliance upon individual judicial caution is undesirable where the process can determine whether one participant is prosecuted and whether his evidence is used against others.

B. Prosecutorial influence and custody

The prosecution’s investigative knowledge makes its participation necessary, but its control also creates a risk that it will select a witness who supports its preferred theory. The court should examine whether the proposed approver offers genuinely material evidence, whether his account is independently verifiable and whether the prosecution already has sufficient evidence.

Somveer Singh v. State of Haryana illustrates that a pardon request may be refused where sufficient evidence already exists and the application appears to confer an indirect benefit rather than assist the prosecution. The Punjab and Haryana High Court upheld the trial court’s refusal on exactly that reasoning, observing that the purpose of pardon is to help the prosecution prove its case where there is otherwise some dearth of evidence.³² The case supports a necessity-based approach.

Custody requires special scrutiny. A person may accept pardon because detention, fear and uncertainty make the benefit appear unavoidable. The Magistrate should inquire into detention, access to counsel, the timing of cooperation and communications with investigators. After acceptance, custody should be reviewed periodically and investigators should not control the witness’s continuing account.

C. Complete disclosure and corroboration

The defence must receive every material term of the pardon arrangement. This includes promises relating to immunity, release, protection, reduction of charges or treatment in another proceeding, whether written or oral. Prior statements must also be disclosed so that omissions and improvements can be tested.

Courts should treat corroboration as a real safeguard. Independent evidence should connect the accused with material particulars of the offence. It may be documentary, electronic, forensic or testimonial, but it should not consist merely of the approver repeating his story. Nor should the approver’s own earlier statement be treated as independent support.

Corroboration need not prove the entire case without the approver. It must, however, provide reasonable assurance that the testimony is not a self-serving invention. The judgment should identify each important link and the evidence supporting it.

D. Procedural recommendations

First, the Magistrate should conduct a formal voluntariness inquiry. The proposed approver should be informed of the conditional nature of pardon, the duty of complete truthfulness, cross-examination, the possibility of prosecution for breach and the right to refuse. Understanding should be recorded in the person’s own language.

Second, meaningful legal assistance should be available before acceptance. Counsel can explain the consequences without replacing the person’s independent decision. Legal assistance is especially important where the person is detained, uneducated or unfamiliar with criminal procedure.

Third, the court should give a specific reasoned order identifying the necessity of the evidence, the proposed approver’s knowledge and the expected disclosure. The application, objections, statements and order should be preserved in full, preferably through audio-video recording.

Fourth, all material promises and communications should be disclosed to the accused. The prosecution should not rely on an informal bargain outside the record. The court should also ensure timely disclosure of the approver’s earlier accounts.

Fifth, custody should be periodically reviewed. Detention must serve a legitimate protective purpose and cannot be used to compel acceptance or preserve a prosecution narrative.

Sixth, conviction substantially based on approver testimony should ordinarily require independent evidence connecting the accused with material particulars. The court should explain the corroboration in its judgment.

Seventh, breach proceedings should require notice, an opportunity to explain and a reasoned finding of deliberate and material concealment. Minor inconsistencies and genuine memory failure should be distinguished from intentional falsehood.

E. Findings

The study finds that tendering pardon remains necessary for prosecuting serious offences committed secretly or by organised groups. Its primary purpose is to secure evidence against other offenders, not to reward the accomplice. The BNSS substantially preserves the CrPC framework, making earlier precedents relevant, although the BSA has not reproduced Section 133 of the Indian Evidence Act word for word.

Approver evidence is legally admissible but inherently suspect because the witness is a participant and beneficiary. Corroboration must be independent, material and connected with the accused; evidence merely proving that an offence occurred is inadequate. The existing framework does not sufficiently regulate voluntariness, disclosure, informal benefits or custody. Judicial independence and legislative clarification are therefore essential.

F. Proportionality and implementation

The decision to tender pardon should also satisfy a principle of proportionality. The court should compare the investigative advantage expected from the proposed testimony with the cost of granting immunity to a participant. This comparison does not require the judge to calculate punishment in advance, but it does require attention to the person’s relative responsibility, the seriousness of the offence and the availability of alternative evidence. Immunity is more defensible where the person can expose a wider criminal design that cannot otherwise be proved. It is less defensible where the person’s evidence is cumulative, vague or directed only against a particular rival.

The court should formulate the disclosure obligation with sufficient precision. A direction requiring the approver to disclose “everything” may be legally correct but practically uncertain. The order should identify the transaction, relevant period, persons, places, communications, financial arrangements and documents within the expected field of disclosure, without limiting the statutory duty. This will help determine later whether an omission was material and deliberate.

Implementation should be supervised through a standard judicial checklist. The record should show the person’s identity and status, the offence, the explanation of conditions, access to counsel, absence of improper inducement, the terms of every assurance, the approver’s acceptance and the arrangements for custody and protection. A standard record would improve consistency without preventing the court from adapting the inquiry to the facts of the case.

Finally, the trial judge should evaluate the approver’s testimony in stages. First, the court should identify facts based on personal knowledge. Second, it should separate admissions against the approver from allegations against others. Third, it should locate independent corroboration for each material allegation. Fourth, it should consider whether the remaining evidence proves guilt beyond reasonable doubt. This method prevents a persuasive narrative from replacing the disciplined assessment required by criminal law.

Conclusion

Tendering pardon to an accomplice remains an important but controversial mechanism of Indian criminal procedure. Secret offences, conspiracies and organised crimes may be impossible to prove through external evidence alone. A participant may possess the only knowledge capable of explaining the agreement, hierarchy, instructions, division of tasks and movement of proceeds. The State therefore requires a lawful method of obtaining insider evidence.

The mechanism is controversial because the approver is not impartial. He receives conditional protection and may minimise his own responsibility, exaggerate another person’s role or tailor his account to the prosecution’s case. Legal admissibility is consequently different from practical reliability. The fact that a person has accepted pardon cannot itself establish that his testimony is truthful.

Sections 343 to 345 of the BNSS retain the essential CrPC scheme. A competent Magistrate may tender pardon during investigation, inquiry or trial, and the court to which a case has been committed may do so after commitment and before judgment, in each case to a person concerned with a serious offence and for the purpose of obtaining evidence, subject to full and true disclosure. Reasons must be recorded, the approver must be examined according to the prescribed procedure, and breach may lead to prosecution. The BSA continues to recognise an accomplice as competent while warning courts against relying upon his uncorroborated account.

The constitutional framework strengthens these safeguards. Articles 14 and 21 require non-arbitrary and fair procedure. Article 20(3) requires voluntariness where an accused chooses to disclose his involvement. The co-accused must receive the material necessary to expose the approver’s motive and conduct meaningful cross-examination.

The prosecution’s request cannot bind the court. The judge must independently consider necessity, materiality, suitability and voluntariness. Pardon should be refused where evidence is already sufficient or where the request appears designed merely to favour one participant. Selection should rest on rational evidentiary grounds rather than influence or convenience.

The most important reform is a formal voluntariness inquiry. The Magistrate should explain the conditional benefit, the duty of complete disclosure, the right to refuse, cross-examination and the consequences of breach. Legal assistance, a complete record and audio-video preservation would strengthen reliability. Custody should be periodically reviewed so that detention does not become indirect compulsion.

Complete disclosure is equally essential. The defence should receive the pardon order, all earlier statements and every material assurance concerning immunity, release, protection or other treatment. Without that information, the defence cannot show why the witness may favour the prosecution. Corroboration should ordinarily consist of independent evidence connecting the accused with material particulars of the offence. The approver’s own statements should not corroborate themselves.

The mechanism should neither be abolished nor converted into an investigative shortcut. It should remain exceptional, transparent and judicially supervised. Its legitimacy depends upon voluntary acceptance, reasoned discretion, complete disclosure, meaningful cross-examination, reliable records and independent corroboration. Criminal justice seeks not conviction at any cost but a fair determination of guilt. Approver evidence can reveal the truth, but only disciplined judicial scrutiny can prevent it from becoming the source of injustice.

*****

Footnotes

1. Bharatiya Nagarik Suraksha Sanhita, No. 46 of 2023, § 343 (India).

2. Id. §§ 343–45; Code of Criminal Procedure, No. 2 of 1974, §§ 306–08 (India) (repealed with effect from July 1, 2024).

3. Bharatiya Sakshya Adhiniyam, No. 47 of 2023, §§ 119, 138 (India).

4. Bharatiya Sakshya Adhiniyam § 138.

5. Bharatiya Nagarik Suraksha Sanhita § 345.

6. R. v. Baskerville, [1916] 2 K.B. 658, 667–68 (Eng.).

7. Suresh Chandra Bahri v. State of Bihar, 1995 Supp. (1) S.C.C. 80, 96–97.

8. Bharatiya Nagarik Suraksha Sanhita § 343(1).

9. Id. § 343(2).

10. Id. §§ 6, 343(1); cf. Code of Criminal Procedure § 306(1).

11. Bharatiya Nagarik Suraksha Sanhita § 343(3).

12. Lt. Commander Pascal Fernandes v. State of Maharashtra, A.I.R. 1968 S.C. 594, 595–96.

13. State of Maharashtra v. Abu Salem Abdul Kayyum Ansari, (2010) 10 S.C.C. 179, 192–200.

14. Bharatiya Nagarik Suraksha Sanhita § 343(4)(b).

15. India Const. art. 21; Maneka Gandhi v. Union of India, (1978) 1 S.C.C. 248, 283–84.

16. Bharatiya Nagarik Suraksha Sanhita § 344.

17. C.B.I. v. Ashok Kumar Aggarwal, (2013) 15 S.C.C. 222, 229–30.

18. Bharatiya Nagarik Suraksha Sanhita § 345(1).

19. Id. § 345(3), (5).

20. Bharatiya Sakshya Adhiniyam § 138.

21. Indian Evidence Act, No. 1 of 1872, § 133 (India) (repealed) (a conviction “is not illegal merely because it proceeds upon the uncorroborated testimony of an accomplice”); cf. Bharatiya Sakshya Adhiniyam § 138 (a conviction “is not illegal if it proceeds upon the corroborated testimony of an accomplice”).

22. Bharatiya Sakshya Adhiniyam § 119, illus. (b).

23. Sarwan Singh Rattan Singh v. State of Punjab, A.I.R. 1957 S.C. 637, 641–42.

24. Zahira Habibulla H. Sheikh v. State of Gujarat, (2004) 4 S.C.C. 158, 187–89.

25. C.B.I. v. Ashok Kumar Aggarwal, supra note 17, at 229–30.

26. Lt. Commander Pascal Fernandes, supra note 12, at 595–96.

27. Abu Salem Abdul Kayyum Ansari, supra note 13, at 192–200.

28. State through C.B.I., Chennai v. V. Arul Kumar, (2016) 11 S.C.C. 733, 741–45.

29. India Const. art. 21; Maneka Gandhi, supra note 15, at 283–84.

30. India Const. art. 14.

31. India Const. art. 20(3).

32. Somveer Singh v. State of Haryana, 2023:PHHC:080791, ¶¶ 18–25 (Punjab & Haryana High Ct. June 2, 2023).