FPR · First Precedent Research
2026 FPR EPR 7 · https://firstprecedent.org/principles/quality-not-quantity-of-evidence
PR-0079Principles · Evidence · Editorial draft · 2026 FPR EPR 7
FPR Editors31 October 2026 · 4,386 words
Quality, Not Quantity, of Evidence
Under section 134 of the Indian Evidence Act, 1872, now section 139 of the Bharatiya Sakshya Adhiniyam, 2023, no particular number of witnesses is required for the proof of any fact; evidence is weighed and not counted, so a court may act on a single witness who is wholly reliable and must seek corroboration from a witness who is neither wholly reliable nor wholly unreliable, while a relationship to the victim does not by itself make a witness untrustworthy.
Abstract
Indian law does not count witnesses. This entry sets out section 134 of the Indian Evidence Act, 1872 and the classification of testimony in Vadivelu Thevar v. State of Madras; later statements of the rule in Lallu Manjhi, Namdeo and Bipin Kumar Mondal; the treatment of related witnesses in Dalip Singh v. State of Punjab; the weight of injured witnesses in Abdul Sayeed v. State of M.P.; the testimony of a prosecutrix in State of Punjab v. Gurmit Singh; the demand for a high quality of sole testimony in Rai Sandeep; the rule of prudence in cases of many accused associated with Masalti v. State of U.P.; and the effect of not examining a material witness in Habeeb Mohammad v. State of Hyderabad.
I. The principle
Section 134
1Section 134 of the Indian Evidence Act, 1872 provides in a single sentence: “No particular number of witnesses shall in any case be required for the proof of any fact.” Section 139 of the Bharatiya Sakshya Adhiniyam, 2023 carries the rule forward. The maxim usually associated with the provision is testes ponderantur, non numerantur, witnesses are to be weighed and not numbered.
2The rule is not self-evident. Some systems have required more than one witness to prove serious matters, and the temptation to think that two witnesses are better than one is natural. The Indian rule rejects that as a matter of law. Bipin Kumar Mondal v. State of West Bengal gives the modern statement:
There is no legal impediment in convicting a person on the sole testimony of a single witness. That is the logic of Section 134 of the Evidence Act, 1872. But if there are doubts about the testimony the courts will insist on corroboration. In fact, it is not the number, the quantity, but the quality that is material. The time-honoured principle is that evidence has to be weighed and not counted. The test is whether the evidence has a ring of truth, is cogent, credible and trustworthy or otherwise.Bipin Kumar Mondal v. State of West Bengal, (2010) 12 SCC 91, AIR 2010 SC 3638 (B.S. Chauhan J.).
3Namdeo v. State of Maharashtra puts it in similar terms: “it is well-settled that it is quality of evidence and not quantity of evidence which is material. Quantity of evidence was never considered to be a test for deciding a criminal trial and the emphasis of Courts is always on quality of evidence.”Namdeo v. State of Maharashtra, (2007) 14 SCC 150 (C.K. Thakker J.). The sole eyewitness in that case was the son of the deceased.
The reason
4There are practical reasons as well as logical ones. Many crimes are witnessed by few people. If the law insisted on two witnesses, a crime seen by one person could not be punished. The Supreme Court in Vadivelu Thevar v. State of Madras also identified another danger: “if courts were to insist on plurality of witnesses in proof of any fact, they will be indirectly encouraging subornation of witnesses.”Vadivelu Thevar v. State of Madras, AIR 1957 SC 614, 1957 SCR 981 (B.P. Sinha J.). A fixed numerical requirement would give a litigant an incentive to supply another witness.
II. The classification
Vadivelu Thevar
5The leading authority is Vadivelu Thevar. The court said that oral testimony in this context may be classified into three categories, “namely: (1) Wholly reliable. (2) Wholly unreliable. (3) Neither wholly reliable nor wholly unreliable.” It then explained how a court approaches each one:
In the first category of proof, the court should have no difficulty in coming to its conclusion either way-it may convict or may acquit on the testimony of a single witness, if it is found to be above reproach or suspicion of interestedness, incompetence or subornation. In the second category, the court, equally has no difficulty in coming to its conclusion. It is in the third category of cases, that the court has to be circumspect and has to look for corroboration in material particulars by reliable testimony, direct or circumstantial.Vadivelu Thevar, AIR 1957 SC 614.
6Lallu Manjhi v. State of Jharkhand restates this: “The Law of Evidence does not require any particular number of witnesses to be examined in proof of a given fact. However, faced with the testimony of a single witness, the Court may classify the oral testimony into three categories”.Lallu Manjhi v. State of Jharkhand, (2003) 2 SCC 401 (R.C. Lahoti J.).
The difficult third category
7The first and second categories are mostly simple. Most real witnesses are in the third. Corroboration is therefore a rule of prudence that works alongside section 134 and not a legal requirement. The rule is about what a careful judge would look for, given the particular witness. It is not an exception to section 134. Corroboration may come from other witnesses, from documents, from medical evidence, from the accused’s conduct, or from objects found. What is required is corroboration on material particulars, and the court should look to the essential facts of the case.
Sterling witnesses
8Rai Sandeep v. State (NCT of Delhi) describes the first category in a demanding way:
In our considered opinion, the ‘sterling witness’ should be of a very high quality and caliber whose version should, therefore, be unassailable. The Court considering the version of such witness should be in a position to accept it for its face value without any hesitation.Rai Sandeep v. State (NCT of Delhi), (2012) 8 SCC 21 (Bench: Swatanter Kumar and Fakkir Mohamed Ibrahim Kalifulla JJ.).
9The test of such a witness is the truthfulness and consistency of the statement “right from the starting point”. Thus the fact that the law permits a conviction on one witness does not make conviction on one witness easy.
Amar Singh
10Amar Singh v. State (NCT of Delhi) shows the same caution in operation. Other eye witnesses, including one who was declared hostile, did not support the prosecution, and it was argued that “it was not safe to rely upon the solitary evidence of the sole remaining eye witness”, whose presence at the scene was said to be doubtful because of his conduct.Amar Singh v. State (NCT of Delhi), (2020) 19 SCC 165, AIR 2020 SC 4894 (Krishna Murari J.). The argument illustrates how the sole-witness rule usually arises. The court asks whether this person’s account is one that can be relied on alone, in light of the conduct and the rest of the record.
III. Related and interested witnesses
Dalip Singh
11A related witness is not a tainted witness. Dalip Singh v. State of Punjab, in the language of Vivian Bose J., corrected this assumption more than seventy years ago:
A witness is normally to be considered independent unless he or she springs from sources which are likely to be tainted and that usually means unless the witness has cause, such as enmity against the accused, to wish to implicate him falsely. Ordinarily, a close relative would be the last to screen the real culprit and falsely implicate an innocent person.Dalip Singh v. State of Punjab, AIR 1953 SC 364, 1954 SCR 145 (Vivian Bose J.).
12The Court acknowledged that feelings may run high, and that where there is personal cause for enmity, “there is a tendency to drag in an innocent person against whom a witness has a grudge along with the guilty”. But a foundation must be laid for that suggestion. It cannot be assumed from the relationship alone. In the Court’s words, the view seemed to persist in counsel’s arguments, and the Court corrected it.
Interested witnesses
13The two concepts are distinct. A relative is simply someone connected to the victim. An interested witness is someone with some reason to want a particular person convicted, such as enmity or a stake in the outcome. Evidence of an interested witness calls for caution. It is not automatically to be rejected. Courts therefore ask two questions: whether there is a real motive to falsely implicate, and whether the account stands up on its own and against other evidence.
IV. Particular witnesses
The injured witness
14An injured witness is generally treated as reliable, for reasons Abdul Sayeed v. State of M.P. gives:
Where a witness to the occurrence has himself been injured in the incident, the testimony of such a witness is generally considered to be very reliable, as he is a witness that comes with a built-in guarantee of his presence at the scene of the crime and is unlikely to spare his actual assailant(s) in order to falsely implicate someone.Abdul Sayeed v. State of M.P., (2010) 10 SCC 259 (B.S. Chauhan J.).
15The passage goes on to say that convincing evidence is required to discredit an injured witness. This does not make the testimony unquestionable. The injury shows presence. The account given still needs to be consistent with the other evidence.
The prosecutrix
16The classic statement in sexual offence cases concerns corroboration. State of Punjab v. Gurmit Singh, (1996) 2 SCC 384, is the authority most often cited. It emphasises that courts should approach the testimony of a victim of sexual assault with sensitivity, that the testimony can be acted on without corroboration if it inspires confidence, and that it should not be rejected on account of minor discrepancies. The trial court in that case had disbelieved the prosecutrix for reasons that included the absence of corroboration, and the Supreme Court rejected that approach. The combination of this case and Rai Sandeep gives the practical rule: corroboration is not a legal necessity, but the testimony must in fact be of a quality that can sustain the conclusion.
V. Many accused
The rule of prudence in Masalti
17A rule of prudence may apply where a large group is charged with an attack, and where it is difficult to be sure of each person’s identification. Masalti v. State of U.P., AIR 1965 SC 202, is regularly cited for the point that a court may require consistent evidence from a certain number of witnesses before convicting a particular member of such a mob. The rule has been stated as a rule of prudence for this special setting. It does not change section 134, and it does not mean that a fixed number is necessary as a matter of law. The source could not be located in full text for this entry, so a reader should consult the report before relying on its terms. The reason for caution is easily understood: in a large and confused event, partisan witnesses may add names, and an inquiry into a particular individual’s participation requires care.
VI. Not examining a witness
The material witness
18Section 134 frees a party from having to call a number of witnesses. It does not free the prosecution from presenting fairly the material evidence. Habeeb Mohammad v. State of Hyderabad is the leading older authority. The officer who had accompanied the accused and had witnessed the occurrence was not examined. The Court held “that the failure to examine him not only led to an adverse inference against the prosecution case but also cast serious reflection on the fairness of the trial.”Habeeb Mohammad v. State of Hyderabad, AIR 1954 SC 51, 1954 SCR 475 (Mehr Chand Mahajan J.), reported headnote.
19Section 114 illustration (g) of the Act allows the court to presume that evidence which could be and is not produced would, if produced, be unfavourable to the person who withholds it. Whether a court draws that inference depends on whether the witness was truly material, whether the party had a reason not to call that person, and whether the rest of the evidence is otherwise adequate. The prosecution is not bound to call every witness, and a failure to call an independent witness is not fatal where the evidence called is reliable. But unexplained omission of a material witness is a serious matter.
Independent witnesses
20Courts are aware that people are often unwilling to testify, particularly in villages and in factional disputes, and a failure to produce independent witnesses is not by itself decisive. A court considers whether independent witnesses were available and why they were not produced.
VII. Beyond crime
21Section 134 is general, and it applies in civil cases as well. It must however be read with statutes requiring particular formalities. Section 63 of the Indian Succession Act, 1925 requires a will to be attested, and section 68 of the Evidence Act requires at least one attesting witness to be called to prove a document required to be attested. These specific rules are not in conflict with section 134, since they are rules about particular documents.
VIII. Practice
For the prosecution
22Do not simply call more witnesses, particularly if their evidence adds nothing. Choose witnesses who were placed to see and who can be relied on. Show why any witness the defence would expect to see is absent. Find corroboration on material particulars where the principal witness is in the third category.
For the defence
23Do not stop at the number of witnesses. Work out what category the witness falls into, and show why that witness is not wholly reliable. Establish any motive to falsely implicate. Note any material witness who was withheld, and ask the court to draw the inference. Bear in mind that the fact of a relationship with the victim will not by itself help.
IX. How courts assess a witness
The usual considerations
24Calling evidence “weighed and not counted” does not by itself explain how the weighing is done. The reported decisions show a fairly consistent list of questions. Was the witness in a position to see what is described? Is the account consistent with itself, and with earlier statements? Does it fit the medical evidence and the physical facts? Did the witness have any reason to lie? Was the account given promptly, or did it emerge after an unexplained delay? And how did the witness stand up to cross-examination?
25None of these is decisive on its own. Earlier cases have said that minor contradictions, which do not go to the root of the case, should be expected and may even show that the witness has not been coached. A witness who tells a story in exactly the same words each time may be less convincing than one with small variations. The test is whether a discrepancy touches something that matters.
The reaction of witnesses
26A common line of attack is that a witness did not behave as a witness “should” have behaved. The witness did not run to help, or did not raise an alarm, or was silent for some time. The Supreme Court has warned against reasoning from an assumed normal response. Appabhai v. State of Gujarat says:
The Court, however, must bear in mind that witnesses to a serious crime may not react in a normal manner. Nor do they react uniformly. The horror stricken witnesses at a dastardly crime or an act of egregious nature may react differently. Their, course of conduct may not be of ordinary type in the normal circumstances. The Court, therefore, cannot reject their evidence merely because they have behaved or reacted in an unusual manner. In Rana Pratap and Ors. v. State of Haryana 1988 (3) S.C.C. 327 O. Chinnappa Reddy J. speaking for this Court succinctly set out what might be the behaviour of different persons witnessing the same incident. The learned Judge observed; (at p. 330). Every person who witnesses a murder reacts in his own way. Some are stunned, become speechless and stand rooted to the spot. Some become hysteric and start wailing. Some start shouting for help. Others run away to keep themselves as far removed from the spot as possible. Yet others rush to the rescue of the victim, even going to the extent of counter-attacking the assailants. Every one reacts in his own special way. There is no set rule of natural reaction.Appabhai v. State of Gujarat, AIR 1988 SC 696, quoting Rana Pratap v. State of Haryana.
27The passage is often relied on by prosecutors, and it is sound. But it does not remove the need for the conduct to make sense. If a witness claims to have seen a killing and then attends to ordinary work for two days without telling anyone, the court will want an explanation. The point is that there is no single expected reaction, and that an unusual reaction does not by itself destroy a witness. It is a factor that must be considered with the rest.
X. Independent witnesses
Appabhai on their absence
28The same case deals with the failure to produce independent witnesses. The incident took place at a busy bus stand, and no bystander was called. It was argued that the prosecution case could not be accepted without one. The Court answered:
It is no doubt true that the prosecution has not been able to produce any independent witness to the incident that took place at the bus stand. There must have been several of such witnesses. But the prosecution case cannot be thrown out or doubted on that ground alone. Experience reminds us that civilized people are generally insensitive when a crime is committed even in their presence. They withdraw both from the victim and the vigilante. They keep themselves away from the Court unless it is inevitable. They think that crime like civil dispute is between two individuals or parties and they should not involve themselves. This kind of apathy of the general public is indeed unfortunate, but it is there everywhere whether in village life, towns or cities. One cannot ignore this handicap with which the investigating agency has to discharge its duties. The court, therefore, instead of doubting the prosecution case for want of independent witness must consider the broad spectrum of the prosecution version and then search for the nugget of truth with due regard to probability if any, suggested by the accused.Appabhai, AIR 1988 SC 696.
29The reasoning is practical. People avoid courts, and the law should not insist on evidence that the investigating agency cannot compel in reality. Yet the absence of independent witnesses can still matter. It is a factor to be weighed. If it would have been easy to obtain such evidence, and if the only witnesses called have a clear motive to implicate the accused, a court may well be less ready to accept the case. The Habeeb Mohammad principle on material witnesses, considered above, represents the more serious kind of failure, where a witness who was certainly present and is under the prosecution’s control has been held back.
Animosity as a double-edged sword
30The defence often points to enmity between the parties. State of U.P. v. Kishan Chand gives the usual reply:
By now, it is well-settled principle of law that animosity is a double-edged sword. It cuts both sides. It could be a ground for false implication and it could also be a ground for assault. Just because the witnesses are related to the deceased would be no ground to discard their testimony, if otherwise their testimony inspires confidence. In the given facts of the present case they are but natural witnesses. We have no reason to disbelieve their testimony. Similarly, being the relatives, it would be their endeavour to see that the real culprits are punished and normally they would not implicate wrong persons to the crime, so as to allow the real culprits to escape unpunished.State of U.P. v. Kishan Chand, (2004) 7 SCC 629, AIR 2004 SC 4671 (B.N. Agrawal and H.K. Sema JJ.).
31The first sentence is often quoted. Enmity is a possible reason for a false accusation, and it is also a possible motive for the crime. So the presence of enmity is not itself enough to reject a witness, and it cannot be used only for one side. The Court’s words make clear that the matter comes down to whether the testimony inspires confidence.
32The Court went on to say that “The testimony of an injured witness has its own relevance and efficacy. The fact that the witnesses sustained injuries at the time and place of occurrence lends support to their testimony that the witnesses were present during the occurrence.” That is the same reasoning as Abdul Sayeed. It concerns presence, and not necessarily the accuracy of each detail in the account.
Natural witnesses
33Kishan Chand described the relatives in that case as “natural witnesses”. The phrase is useful. A natural witness is someone whose presence at the scene is to be expected in the ordinary course. A family member at home at the time of a crime in the home is a natural witness. A stranger who happens to pass by a lonely place at an odd hour, and who has no clear reason to be there, may be described as a chance witness. Courts look more closely at the latter. They ask why that person was there, and whether the person’s presence is explained. The question is one of probability, and a chance witness may be fully truthful.
XI. The prosecutrix after Gurmit Singh
The later statements
34The rule in Gurmit Singh has been restated often. Phool Singh v. State of Madhya Pradesh records the State’s submission, drawing on State (NCT of Delhi) v. Pankaj Chaudhary, that “conviction can be sustained on the sole testimony of the prosecutrix if it inspires confidence and that there is no rule of law or practice that the evidence of the prosecutrix cannot be relied upon without corroboration.”Phool Singh v. State of Madhya Pradesh (Supreme Court, 1 December 2021) (M.R. Shah and Sanjiv Khanna JJ.), recording the submission that relied on State (NCT of Delhi) v. Pankaj Chaudhary, (2019) 11 SCC 575. The conviction in that case was confirmed and the appeal dismissed. The absence of external or internal injuries, in an examination three days after the incident, was the principal attack made on the evidence.
35Santosh Prasad v. State of Bihar went the other way on its facts. The case for the defence was that the medical evidence did not support the prosecution, that the deposition was “full of material contradictions”, that there was an existing dispute between the families, and that no independent witnesses had been examined. The Court acquitted, saying that the version of the prosecutrix “cannot be taken as a gospel truth at face value and in the absence of any other supporting evidence, there is no scope to sustain the conviction”.Santosh Prasad v. State of Bihar, (2020) 3 SCC 443, AIR 2020 SC 985 (M.R. Shah and Ashok Bhushan JJ.). The two decisions, by benches with a judge in common, illustrate how far the result depends on the record.
36These cases should be read together. Neither states a different rule. Corroboration is not a requirement of law. The testimony must inspire confidence. Whether it does so depends on how it stands against the other evidence. A defence advocate should not suggest that the absence of injuries or of corroboration is enough on its own to create doubt. A prosecutor should not suggest that the evidence must be believed because of the nature of the offence. The Court’s rule on the sensitivity with which such evidence is to be approached does not relieve the court of its duty to test it.
The statutory presumptions
37Parliament has also added to this field. Section 114A of the Evidence Act, inserted in 1983, gives rise to a presumption as to absence of consent in certain prosecutions, and section 53A was added in 2013 to limit the use of evidence of the character or previous sexual experience of a victim. Those provisions are outside section 134, but they bear on how such evidence is assessed, and they are carried forward in the Bharatiya Sakshya Adhiniyam, 2023 in provisions that should be consulted under the current numbering.
XII. Mobs and Masalti in context
38Kishan Chand also shows how Masalti is cited in unlawful assembly cases, for a separate proposition. It quotes the Constitution Bench on what has to be proved against a person alleged to be a member of an unlawful assembly, and the Court says:
when the conviction is recorded with the aid of Section 149, relevant question to be examined by the court is whether the accused was a member of unlawful assembly and not whether he actually took active part in the crime or not.Kishan Chand, (2004) 7 SCC 629, referring to Masalti v. State of U.P., 1964 (8) SCR 133.
39This matters for the rule of prudence described above. Because membership of the assembly is enough for liability under section 149, the evidence must show membership with some care. A witness who names a long list of people as being present, in an atmosphere of faction, is giving evidence that could convict each of them for acts done by others. That is why the courts look for more than a single witness on this point in a large and confused incident. The caution attaches to the risk that a name has been added. It is not a general rule that any given number of witnesses is needed.
XIII. Identification evidence
40A related problem concerns evidence that a single witness identified the accused. Identification by a single witness can be enough, as with any other fact. But courts look at the opportunity to observe, the lighting, the time available, the description given at first, how long it was before the identification, and whether the witness knew the person earlier. An identification in the course of a sudden and frightening incident needs more care than an identification of a neighbour. Section 134 does not change those considerations, since they go to weight.
XIV. Civil cases and documents
41In a civil case the same principle operates in a less dramatic way. A single witness can prove a fact, and the court may prefer that witness to several others on the opposite side. But the civil standard of proof differs, and a party who has a document in its possession and who relies on an oral account instead may face the inference in section 114(g). A plaintiff who proves title or a contract through one witness who is believed has done enough, unless a particular statute requires more.
42The rule applies with the same force to expert evidence. There is no requirement for more than one expert, and a court is not obliged to prefer the side with the greater number of experts. The question is which opinion is better reasoned, and which rests on the more reliable data.
XV. How the principle operates at appeal
43The principle also affects appeals. Since an assessment of quality depends largely on what the witness said and how it was said, an appellate court is cautious in disturbing the findings of the court that saw the witness. A trial judge’s opinion of a witness, where it is based on demeanour and is recorded, carries real weight. But an appellate court can and should test whether the reasons given for accepting or rejecting a witness actually hold up on the record. Kishan Chand is a useful example, in which the Supreme Court set aside an acquittal, and restored the conviction recorded by the trial court, where the High Court “did not assign any reason much less ostensible reason for discarding the testimony of the eyewitnesses”.
44A judgment that merely says a witness is or is not reliable, without giving reasons, does not show that the evidence has been weighed. So a court should say why it believes or disbelieves a witness. A party who wants to challenge such a finding should concentrate on the reasons and on what the court failed to consider, rather than on a bare count of how many witnesses supported each side.
XVI. The position today
45No particular number of witnesses is required for the proof of any fact, and a court may convict or give judgment on the testimony of a single witness who is wholly reliable. A witness who is wholly unreliable should be rejected. The court must be circumspect with a witness who is neither, and look for corroboration in material particulars. A relative is not for that reason an unreliable witness, though the court considers whether there is enmity or other cause to implicate falsely. An injured witness comes with a built-in guarantee of presence. Rules of prudence may apply to the testimony of a prosecutrix, though without any requirement of corroboration, and to identification in cases of many accused. A party’s failure to examine a material witness may lead to an adverse inference. Evidence is to be weighed and not counted.
History
- Published · 31 October 2026
© 2026 First Precedent Research. Free to share and adapt with credit, for non-commercial use, under CC BY-NC 4.0. A published piece is changed only to correct an error, and each correction is listed under History. If the law changes, we add a linked afterword. Nothing on FPR is legal advice.