JUDGMENT Hari Swarup, J. 1. This reference has been occasioned because of different standards being applied to the testing of the veracity of identifying witnesses by different Judges of the Court. When the appeals came up for hearing before a learned single Judge, it was argued that the identification evidence of a witness who at the test identification parade commits even 33% mistake should be discarded as unreliable. Learned single Judge finding that different Judges had taken different views referred the following question of law for the opinion of the Division Bench : "What should be the minimum percentage of correct identification which should be considered to be good identification ?" 2. In criminal cases where a miscreant is not identified by name, he may identified by the witnesses on the basis of his appearance or voice. In such a case he has to be identified by the witness in the court at the time he gives testimony. The authenticity of the evidence of the witness by which he points out a particular accused as the participant in the crime is normally tested by the court from his performance at the test identification parade held during the course of investigation. The courts have always found it prudent to test the witness's testimony in court by his conduct at the test parade. At the investigation stage when a parade is held and the suspect is mixed up with persons who admittedly could not have been present at the time of occurrence the witness gets opportunity of picking out the person whom he alleges to be the offender. He may at such a parade pick out that man and no other, he may pick out that man and also some others, or he may pick out not the suspect but some one who was admittedly not at the scene of occurrence. The courts have taken these parades into consideration in judging the authenticity of the identification evidence of the witness at the trial. Certain cases have been cited before us to show how the Judges have reacted to the performance of witnesses at identification parades. The courts have taken these parades into consideration in judging the authenticity of the identification evidence of the witness at the trial. Certain cases have been cited before us to show how the Judges have reacted to the performance of witnesses at identification parades. In Emperor v. Debi Char an alias Debi Prasad, 1942 AWR 293, (Ganga Nath and Yorke, JJ.), it was held that if a person makes I correct identification and commits I mistake, i. e., his performance is 50 per cent, he would not be a witness worthy of reliance for the purposes of identification. But, if a person makes 3 correct identifications and commits two mistakes, i. e., his performance is 60 per cent, he would be a good witness whose testimony at the trial about identification can be relied upon. In State v. Wahid Bux, 1952 AWR 512 (Dayal and Agrawala, JJ.), it was held that where the result at the identification is 75% correct the witness would be reliable. In Munshi v. State, AIR 1968 All. 345 = 1967 AWR 689, (Gangeshwar Prasad, J.), the learned single Judge held that a witness was not reliable if he correctly identified 2 and made one mistake, which means that 66 per cent correct identification was not reliable. In Soorajpal v. State, 1977 ACC 68, (B. N. Katju, J.), the learned Judge took the view that 3 correct and 1 wrong was not reliable. Similar view was taken in Iqbal v. State, 1976 ACrJ 7 (S. K. Kaul, J.). In Ali Sher v. State, 1979 LLJ 8 (Prem Prakash, J.), a witness who had correctly identified 4 suspects but made 1 mistake was held to be class-2 witness. In none of these cases we find any reason given for holding why a particular percentage of correct identification makes the witness reliable and why the other percentage makes him unreliable. 3. Besides these cases we have been referred to a judgment of the Supreme Court in Vaikuntam Chandrappa v. State of Andhra Pradesh, AIR 1960 SC 1340 . In that case the Supreme Court relied on the evidence of a witness who had correctly identified two accused and made one mistake ; which means that 66 per cent correct identification at the identification parade was held sufficient for making the witness reliable for the purposes of identification in court. 4. In that case the Supreme Court relied on the evidence of a witness who had correctly identified two accused and made one mistake ; which means that 66 per cent correct identification at the identification parade was held sufficient for making the witness reliable for the purposes of identification in court. 4. In a case where the witness identifies an accused, he gives the evidence of identification along with other evidence. We are here concerned only with that part of his testimony which relates to the identification of the accused. The relevance of the prior identification parade was accepted by the Supreme Court in Budhsen v. State of U.P., AIR 1970 SC 1321 , wherein it was observed : "As a general rule, the substantive evidence of a witness is a statement made in court. The evidence of mere identification of the accused person at the trial for the first time is from its very nature inherently of a weak character. The evidence in order to carry conviction should ordinarily clarify as to how and under what circumstances he came to pick out the particular accused person and the details of the part which the accused played in the crime in question with reasonable particularity. The purpose of a prior test identification, therefore, seems to be to test and strengthen the trustworthiness of that evidence. It is accordingly considered a safe rule of prudence to generally look for corroboration of the sworn testimony of witnesses in court as to the identity of the accused who are strangers to them, in the form of earlier identification proceeding." In state of U. P. v. Boota Singh, AIR 1978 SC 1770 , the aforesaid observation was reiterated and it was observed that where the witness correctly identifies the accused at a T. I. parade held by a Magistrate after observing all the essential formalities and taking the necessary precautions and then identifies the accused also in court the evidence of identification can be believed unless the evidence of witness suffers from some other infirmity. 5. Sri N. R. Kashyap, learned Government Advocate urged that test parade identification was not evidence within the meaning of Evidence Act. Reference has been made by him to Section 9 of the Evidence Act. He is right that the testification done at the identification parade is not evidence because that is only a proceeding at the investigation stage. 5. Sri N. R. Kashyap, learned Government Advocate urged that test parade identification was not evidence within the meaning of Evidence Act. Reference has been made by him to Section 9 of the Evidence Act. He is right that the testification done at the identification parade is not evidence because that is only a proceeding at the investigation stage. The evidence is what the witness gives before the court. The conduct of the witness at the identification parade, however, becomes relevant and that conduct is proved in court by the evidence of the Magistrate who conducts the identification test. It is his evidence that is evidence and is utilizable for the appreciation of the evidence of the witness who testifies in court that a particular accused was the person who had participated in the crime. 6. The value of test identification is for the purpose of testing the veracity of a witness. A person's conduct at a test parade in identifying wrong person is not deemed fatal because of the mental processes concerning memory which an individual goes through between the time of the incident and the time of identification. False recognition at the time of identification parade does not mean that the man has such a bad memory that his recognition even of the suspect becomes unreliable. The recognition of things and individuals by an individual depends upon various cues. With the lapse of time the cues go on getting reduced but a person is recognised even on the basis of some of the cues. According to Norman L. Munn: "So-called false recognition is another example of response to reduced cues. We may "recognize" a person as our friend because of some similarity to the friend, such as hair-color, walk, build, or dress. Some aspect of former stimulation involving our friend leads us to recall him, and at the same time to identify the present person with the one recalled. The feeling that one has been in a certain place before or that he has done or said something before, even though he knows that this is the first time, has a similar basis. It is often referred to as the dejavu experience. Something in the present situation or present behavior may be identical with, or very much like, something that occurred previously. It is often referred to as the dejavu experience. Something in the present situation or present behavior may be identical with, or very much like, something that occurred previously. This present aspect of former stimulation or activity leads us to recall the original experience and we incorrectly identify it with the present one." Introduction to Psychology, by Norman L. Munn, Bowdoin College, page 321." It is this principle on the basis of which the courts have compared the mistakes with the correct recognitions done by a witness : "Any activity which produces a poor impression obviously yields poor retention. The poorly motivated subject learns little, and, when tested for retention, later, retains little." (From Introduction to Psychology, by Norman L. Munn, Bowdoin College, page 321). 7. It is the retention of the witness of the image of persons that he gets at the time of occurrence which is tested at the time of identification parade. Memory of a person depends on the impact of the incident on the mind and the intensity of the picture on his mental screen. After the incident takes place the memory goes into slumber and it revives when the picture is reproduced. When a person is mixed up with others the witness works on the reduced cues and sometimes commits the mistakes. Mistakes in identification are due to the fact that in the process of restructuring the picture, the mind falters in making a correct picture, particularly when the intensity of the original picture on the mind was not strong enough. When the mistakes are too many the courts do not consider it proper to rely on the witness, but when the mistakes are not many the witness is relied upon. 8. It is not possible to lay down any rule of thumb for determining whether a particular witness can be relied upon or not. IT is here that the observation of Oliver Wendell Holmes becomes relevant : "The life of law has not been logic ; it has been experience." (The Common Law and Collected Legal Papers). It is why we find that the various Judges have applied different tests for relying or not relying on a particular witness. It is experience of each individual Judge that has mattered in the cases that we have cited above. It is why we find that the various Judges have applied different tests for relying or not relying on a particular witness. It is experience of each individual Judge that has mattered in the cases that we have cited above. As we have already seen no judgment has given any reason, rightly so, because the experience depends not on logic but on the nature and probability of human behaviour. When different courts experience differently, it is the experience of the highest court of the land that becomes relevant. What the Supreme Court says has the binding effect and has to be followed by all courts in the country. If the Supreme Court relies on a witness who at the test parade points out two persons rightly and one wrongly, that is, gives a 66 per cent correct result, then no court can say that such a witness will not be relied upon because he commits 33 per cent mistake. All the judgments which have taken a view contrary to the judgment of the Supreme Court in Vaikuntam Chandrappa v. State of Andhra Pradesh (supra), must he held to have taken a view not warranted by law. It was urged that the Supreme Court decision does not hold that a witness whose performance at the test parade is 66 per cent is reliable as the judgment of the Supreme Court was in respect of a corroborating witness. We do not find any substance in this contention. In Vaikuntam Chandrappa v. State of Andhra Pradesh (supra) the Supreme Court considered the evidence of witnesses along with their performance at the test parades. Against accused nos. 5 and 7 there was the evidence of PW 3 who had named them and of PW 2 who had identified them at the test identification parade. PW 2 had picked out two of these suspects and one wrong person. Thus the result of this witness was 66 per cent. Dealing with the case of accused nos. 5 and 7 the Supreme Court observed in paragraph 8 as follows:- "This leaves the case of accused nos. 5 and 7. They were named by PW 3. They were also identified by PW 2 in court whose performance at the test identification was satisfactory. Dealing with the case of accused nos. 5 and 7 the Supreme Court observed in paragraph 8 as follows:- "This leaves the case of accused nos. 5 and 7. They were named by PW 3. They were also identified by PW 2 in court whose performance at the test identification was satisfactory. So far as these two appellants are concerned, the High Court was right in holding that there was sufficient corroboration of the evidence of PW 3. In these circumstances their appeal must fail." 9. Thus the Supreme Court had relied upon two witnesses for holding accused nos. 5 and 7 guilty-one was PW 3 and the other was PW 2. The evidence of PW 2 was relied upon because he had identified these two persons at the test parade even though he had made one mistake. Corroboration can be only of testimonies which are worthy of credence ; only reliable witnesses can corroborate. The argument that the Supreme Court has not held that 66 percent correct identification makes a reliable witness is fallacious. The Supreme Court has held that PW 2 was a reliable witness and that the evidence of PW 3 and PW 2 taken together was sufficient for upholding the conviction. The test of reliability of a witness is not dependent on the test of its sufficiency for conviction. If a witness is reliable then his testimony can be used both as basic and as corroborative evidence. 10. In the present reference we are not concerned with the sufficiency of the evidence, but only with the reliability of an identifying witness. When a witness identifies a person in court and his testimony on being tested by his performance at the identification parade is found to be reliable, then, the court proceeds to decide if the evidence is sufficient to record the conviction. According to the Supreme Court a witness is reliable if his performance at the test identification parade is 66 per cent correct. The evidence of such a witness can, therefore, be utilised for purposes of proving the identity of the participant in crime. According to the Supreme Court a witness is reliable if his performance at the test identification parade is 66 per cent correct. The evidence of such a witness can, therefore, be utilised for purposes of proving the identity of the participant in crime. In our opinion, the real matter in issue can be more appropriately brought into relief by the reframing of the question in the following form : "What should be the normal range of correct and wrong identification by a witness at the test identification parade to make him a reliable or non- reliable identifying witness at the trial ?" We reframe the question accordingly. 11. In view of the decision of the Supreme Court in Vaikuntam Chandrappa v. State of Andhra Pradesh, (supra), our answer to the question is that a witness whose performance at the test identification parade is 66 per cent or more correct is reliable and a witness whose performance is 33 per cent or less is not reliable. 12. Let the papers be returned along with this opinion to the learned single Judge who may be dealing with the case.