WERNER, J.

The defendant has appealed to this court from a judgment of the Supreme Court at Trial Term entered upon a verdict convicting him of the crime of murder in the first degree. The indictment charges that the defendant, in the borough of The Bronx in the city of New York on the 4th day of August, 1913, shot and killed one Patrick Cotter, and that this was done willfully, feloniously and with malicious intent.

Cotter was a police officer, attached to the sixty-fifth precinct of the police department of the city of New York. Sergeant Quilty of this precinct last saw Cotter alive at about seven o'clock in the evening at the corner of Hughes avenue and One Hundred and Eighty-seventh street. Twenty minutes later Quilty received word that an officer had been killed, and investigation revealed the fact that Cotter was the unfortunate victim. The circumstances under which this officer met his death are not the subject of controversy. It is conceded that he was shot by a man whom he had pursued from Belmont avenue through to One Hundred and Eighty-eighth street, and thence southerly into Hughes avenue to about the middle of the block. At this point the fugitive took refuge behind a tree or pole and was heard to say in Italian: "Stand, if you move I am going to shoot you." This statement was followed by two shots, which proceeded from behind the tree or pole, and Officer Cotter sank to the ground, making the sign of the cross as he fell. Marsell, an undertaker who happened to be driving in Hughes avenue, heard the shots, went to the spot, placed the unconscious officer in his vehicle and took him to Fordham Hospital. The officer died on the way, and the autopsy later made by the coroner's physician disclosed that death was due to internal hemorrhage resulting from a gunshot wound which entered the body just below the left shoulder blade, and proceeded forward, upward and to the right. Of this feature of the case it is enough to say that the cause of death and the identification of the deceased were established with absolute certainty.

The theory of the prosecution was that the defendant was the fugitive who shot Officer Cotter, and that the defendant's flight from Belmont avenue was due to his participation in another and earlier shooting affray at the latter place. In that behalf it appears without dispute that the defendant, his friend Delmonico, and the latter's cousin, Norma Scarinci, had visited a moving picture establishment at or near Belmont avenue and One Hundred and Eighty-seventh street, where Delmonico's wife was employed as a singer. As there was no performance in progress the party proceeded to a neighboring restaurant where they had some wine, and then went to walk on Belmont avenue. In front of number 2463 on this thoroughfare they encountered two Italians designated as "Calabrians." It is conceded that an altercation ensued, in the course of which two shots were fired, but at this point the conflict of testimony begins. A number of the People's witnesses saw the disturbance and heard the shooting, but only two of them, Leopold and Buongiorno, identified the defendant as the person who did the shooting. They testified that the defendant pointed a revolver in the direction of the girl, Scarinci, and that he fired two shots. The story as told for the defense, by Delmonico and Scarinci, is that when their party met the Calabrians, the latter bumped into them; that words ensued; that Delmonico struck one of the Calabrians with a cane, and that the Calabrian then drew a revolver and fired two shots. The narrative of what followed is not clear. According to the witnesses for the prosecution the Calabrians quietly vanished in the direction of One Hundred and Eighty-ninth street and Pelham avenue, and the only person who was pursued through Hughes avenue to One Hundred and Eighty-seventh street and Third avenue was the man who was finally caught at the latter place, and he proved to be the defendant. The defendant was not called as a witness at the trial, but his statement taken by a stenographer from the district attorney's office was read in evidence for the prosecution. From this statement it appears that when Delmonico struck one of the Calabrians the latter drew a razor, whereupon the defendant also drew a razor and then the other Calabrian drew a revolver. The defendant says that after the Calabrian with the revolver had fired two shots he started to run and that he, the defendant, followed in pursuit; but that when he heard the two shots in Hughes avenue he, the defendant, was fully a block north of the place where Officer Cotter was shot. For the purpose of this discussion it it unnecessary to follow this long statement to its conclusion. We have given enough of the substance to indicate that the defendant claims to have been a pursuer of the Calabrian who is said to have done the shooting in Belmont avenue, and that he, the defendant, had no idea that he was being pursued until after the shooting in Hughes avenue.

When the defendant was caught he threw away a razor, which was picked up by a bystander and given to a police officer. No revolver was found in the possession of the defendant. Aside from the testimony of the two witnesses who said they saw him fire two shots in Belmont avenue, there is no direct evidence that he ever had a revolver. There are circumstances, however, from which counsel for the prosecution argues that the jury were justified in finding that the defendant had a revolver and that he shot Officer Cotter. In this connection it appears that Captain Price, of the police force, made an examination of the premises already referred to as numbers 2463 and 2465 Belmont avenue, and there he found a bullet mark on a post, another mark on a brick wall to which the bullet had evidently been deflected, and then the bullet itself which had fallen into a narrow space between the two houses. At the time of the shooting in Belmont avenue, the house spoken of as number 2463 was occupied by a Mrs. Pucci, who heard the shooting but did not see it. On the next day, as she was sweeping the floor, she found what she described as a "capsule," but it was in fact the shell of a revolver cartridge, and still later she discovered a hole, evidently made by a bullet, in one of her windows. The defendant is said to have passed through Lorrilard place, in the course by which he finally reached One Hundred and Eighty-seventh street and Third avenue, and some of the witnesses for the prosecution testified that as the fleeing man passed the house on the northeast corner of Lorrilard place and One Hundred and Eighty-seventh street, he apparently threw something from him, and that this was followed by a crash of breaking glass. A later search of these premises by Police Officer Gerhardt, assisted by Fireman Roe, disclosed a broken window in the rear of the building. No pistol was found there, but in the court yard and some four or five feet from the broken window Gerhardt picked up two empty cartridges and a loaded shell. Roe found another bullet on the sill of the broken window. It further appeared that, while the coroner's physician was performing the autopsy on the body of Officer Cotter, a bullet dropped from the clothing. All these cartridges and bullets, thus found in three different places, were identical in calibre and make, and obviously tended to support the contention of the prosecution that the man who did the shooting in Belmont avenue was the man who in his flight shot Officer Cotter.

This short outline of the principal features of the case suffices to indicate that the only real issue presented by the record was over the identity of the slayer of Officer Cotter, and that was clearly a question of the most profound importance. If all the evidence on that subject had been competent we could not say that the verdict against the defendant is without support. But here we come to the vital consideration that much of the testimony on the subject of identification was worse than worthless, and that which was perhaps of highest probative value was utterly incompetent under rules of evidence which we have followed and enjoined from time immemorial.

The homicide was committed in a section of upper New York known as Little Italy. This accounts for the preponderance of Italian witnesses. A number of these were boys. Among them was one Leopold who witnessed the shooting in Belmont avenue, who followed the fugitive from there to Hughes avenue where he saw the shooting of Officer Cotter, and who later saw the defendant after he had been captured. At the police station Leopold identified the defendant as the man who had done the shooting in both places. At the trial his testimony coincided with his prior statements until it came to the identification of the defendant. At this point the witness hesitated and began to explain how the policeman had forced him to identify the defendant. Under the persistent questioning of the district attorney the witness finally pointed out the defendant as the man who had shot the policeman, and there his direct examination ended. His cross-examination was highly interesting, but much too long for repetition. It is enough to say that he retracted his identification of the defendant, and admitted that he had sworn falsely in the Coroner's Court, but sought to excuse himself by stating that he did this under pressure from members of the police force, who told him how the defendant was dressed and particularly drew his attention to a scar on his forehead. When counsel for the defense had succeeded in utterly annihilating Leopold's identification of the defendant the district attorney again took the witness in hand in an effort to rehabilitate him. In this so-called redirect examination the district attorney was very properly given great latitude, but he elicited nothing new except that the witness claimed to have been also intimidated by various persons who were presumably friends of the defendant. Practically the same course was pursued with the witnesses Buongiorno, Mangino, Tomacelli, Civiletti and others. There were differences of degree and circumstance, but in substance the result was in each instance the same. The particular witness either refused to identify the defendant at all, or retracted his identification after it was made. It is hardly necessary to suggest that this was quite an unsatisfactory condition in respect of the most important question in the case, and the district attorney, realizing the situation, reached out for the only remedy immediately available. He called as a witness Officer Repetto, an Italian and a detective attached to the police force, who had figured somewhat in the investigations that followed the death of Cotter. We reproduce the record of his examination.

"Q. Were you present at any time either in the station house or in the Coroner's Office on the 4th day of August, or within two or three days after the 4th of August, when any persons looked at the defendant?

"A. I was.

"Q. Where and when did you see any person look at the defendant and point him out first?"

Counsel for the defendant interposed an objection which was overruled.

"A. I saw the defendant being pointed out in the cell of the 65th Precinct Station house on the night of August 4th, and I also saw persons picking out the defendant in the Coroner's Court on August 7th in the afternoon.

"Q. And what person pointed to him on that occasion?" Same objection. Overruled.

"A. Salvatore Civiletti.

"Mr. Palmieri: I move to strike out the answer on the ground that this is in contradiction and impeachment of a witness for the prosecution who was already on the stand and testified to the contrary from what this officer did, and it is incompetent, irrelevant and immaterial."

[Motion denied.]

"Q. Do you remember what clothing he was wearing on the 7th of August?

"A. He was wearing a dark blue suit that day.

"Q. And while he was wearing a dark blue suit will you state what persons you saw point him out?"

[Objected to as incompetent, irrelevant and immaterial and objection overruled.]

"A. I saw Santo Mangino, Patsy Tomacelli, Joseph Leopold, Joseph Buongiorno point him out and another witness — I don't at present recall his name.

"The Court: I think I ought to say here to the jury that this testimony is simply literally to affect the three parties who pointed out this person. It means nothing else. It does not go any further than that, but that they simply pointed him out.

"Mr. Palmieri: If that is the ruling of the court, if that is the understanding under which the court is permitting this evidence, I except.

"Q. Did you say anything to either of these persons before he pointed out the defendant, regarding the defendant?

"A. I did. I told each one of them, as they would go into the room and the defendant was standing with a couple of others, to go in and see if they could see the man that they saw fire the shots on August 4th, and if there was anybody in there to point him out. Not to be afraid of anyone there."

Counsel for the defense here made another motion to strike out this testimony on the ground, among others, that it was being used by the district attorney for the purpose of impeaching his own witnesses, and the motion was denied.

Probably no rule of evidence is more generally familiar than the rule that a party may not impeach his own witness, and yet there is none that has caused greater confusion in practice. Like many another common-law rule, its genesis is lost in antiquity and the reason for its existence is largely a matter of philosophical speculation. Professor Wigmore, with his characteristic ability and thoroughness, has traced its origin to the primitive times when there were no witnesses in the modern acceptation of the term but "oath helpers" by whose mere oath the issue of the cause was determined, without regard to anything like testimonial support. From that early time its development is largely a matter of tradition until, as Wigmore says, it began to make its occasional appearance as a general working rule in the early part of the eighteenth century, and toward the end of that century it gained general currency by its application in the trial of Warren Hastings. Since that time its existence has been unquestioned, although in a number of jurisdictions it has been abrogated by statute, as in England, Massachusetts, Indiana, Georgia, Kentucky, Texas and perhaps others. In the effort to apply this rule to an infinite variety of circumstances and conditions, we have reaped the usual harvest of conflicting decisions, and the confusion thus created seems to have been augmented by the attempts of the text writers to bring a semblance of order out of chaos. Wigmore boldly expresses the view that there never was any sound reason for the rule, and that, even if there had been, it is no longer to be tolerated as an impediment to the ascertainment of truth by the most direct and certain methods. There is much to be said in support of Professor Wigmore's argument, as will be seen by reference to chapter twenty-nine of his work, but the case at bar is not one in which we should make a new departure, for a human life is at stake. In such a case we should not change a rule of evidence for the purpose of upholding a conviction that cannot be sustained under our own established decisions. Therefore, without further discussion of the reasons which make for or against the continuance of the rule that a party may not directly impeach his own witness, we proceed to a closer view of what was done in the case at bar, and then to a consideration of the authorities.

The witnesses, upon whom the district attorney relied to identify the defendant as the murderer of Officer Cotter, obviously failed to confirm their previous statements or testimony. The district attorney was manifestly surprised. It was, therefore, proper for the learned trial court to give him great latitude in re-examining these shifting witnesses. The record discloses that this was done to the point of extreme indulgence, but the rulings in this regard may fairly be defended by the peculiar exigencies of the case. The district attorney was permitted to go further, however, in calling Repetto to directly contradict Leopold, Civiletti and the others, upon the subject of identification and for the express purpose of proving that before the trial they had made statements and done acts at variance with their testimony at the trial; in other words, that they identified the defendant in the Coroner's Court. This is not permissible under our decisions. In the early case of People v. Safford (5 Den. 112, 118) the defendant was charged with selling liquors without a license. A witness for the prosecution gave testimony which failed to disclose that the defendant had been guilty of any violation of the law. The district attorney was then permitted to interrogate the witness as to what he had sworn before the grand jury. Over the objection of the defendant the witness testified that before the grand jury he had sworn that he had purchased liquor of the defendant. On appeal it was held substantial error to have admitted this evidence. The court there laid down the rule that "Evidence that contradictory statements have been made by a witness, is only allowable with a view to his impeachment, a ground not open to the party producing the witness." The rule thus stated has been consistently adhered to by our courts. It was followed in Thompson v. Blanchard ( 4 N.Y. 303) where one of the parties was allowed to contradict his witness by another witness who testified that the former, prior to the trial, had made statements that conflicted with his testimony at the trial, and it is very clearly stated by Judge PECKHAM in Becker v. Koch ( 104 N.Y. 394, 401) where he said: "Sometimes rather loose language has been indulged in to the general effect that a party cannot impeach his own witness, but when an examination is made as to the limits of the rule the result will be found to be that it only prohibits this impeachment in three cases, viz.: (1) the calling of witnesses to impeach the general character of the witness; (2), the proof of prior contradictory statements by him; and (3), a contradiction of the witness by another where the only effect is to impeach and not to give any material evidence upon any issue in the case. * * * But all the cases concur in the right of a party to contradict his own witness by calling witnesses to prove a fact (material to the issue) to be otherwise than as sworn to by him, even when the necessary effect is to impeach him."

In Coulter v. American Merchants Un. Ex. Co. ( 56 N.Y. 585) the rule was considered, and it was held that "a party may contradict his own witness as to a fact material in the case, although the effect of the proof may be to discredit him; but he cannot impeach him, although subsequently called as a witness for the adverse party, either by general evidence or by proof of contradictory statements out of court." After having thus decided the real question in the Coulter case the learned judge who wrote the opinion went on to discuss an exception to the rule that was not before the court. In the last paragraph of this opinion he said: "There is a class of cases in which a party who calls a witness has been allowed to show, by his own examination at least, if not by introducing proof by others, that he had previously stated the facts in a different manner; but this seems to stand upon the ground of surprise, as contrary to what the party had a right or had been led to believe he would testify, or of deceit through the influence of the other party. (Citing 1 Greenleaf, § 444 and note 1; Melhuish v. Collier, 15 Ad. El. [N.S.] 878; 69 Eng. Com. Law.) But no such special ground of exception appears in this case."

This italicized sentence of Judge JOHNSON clearly indicates that the exception to which the last quoted paragraph relates was not in the case, and was, therefore, not decided. The district attorney now invokes this exception to the general rule. He claims to have been surprised by the testimony of Leopold, Civiletti and others as to the identification of the defendant, and the claim is probably well founded. In disposing of this point we shall not attempt a discussion of the English decisions which were rendered obsolete by the English Common Law Procedure Act of 1853. It is familiar legal history that this act settled in England the question whether a party had a right to contradict his own witness which had long been a matter of contention in the days of Lord DENMAN and Mr. Justice BULLER; the former valiantly contending for the right, while the latter vigorously opposed it. It is enough for present purposes that we look to our own decisions. In Bullard v. Pearsall ( 53 N.Y. 230, 231) the question was whether the trial court properly excluded the testimony of a witness called by the plaintiff for the sole purpose of contradicting another plaintiff's witness by showing that prior to the trial the latter had made statements respecting a date of importance, in which he contradicted his sworn testimony. In discussing this question for the court Judge RAPALLO said: "Where a witness disappoints the party calling him by testifying contrary to the expectations and wishes of such party, it is a conceded rule that the latter shall not, for the purpose of relieving himself from the effect of such evidence, be permitted to prove that the witness is a person of bad character and unworthy of belief. There is also a great weight of authority sustaining the position that under such circumstances the party calling the witness should not be allowed to prove that he has on other occasions made statements inconsistent with his testimony at the trial, when the sole object of such proof is to discredit the witness. * * * The further question has frequently arisen whether the party calling the witness should, upon being taken by surprise by unexpected testimony, be permitted to interrogate the witness in respect to his own previous declarations, inconsistent with his evidence. Upon this point there is considerable conflict in the authorities. We are of opinion that such questions may be asked of the witness for the purpose of proving his recollection, recalling to his mind the statements he has previously made, and drawing out an explanation of his apparent inconsistency. This course of examination may result in satisfying the witness that he has fallen into error and that his original statements were correct, and it is calculated to elicit the truth. It is also proper for the purpose of showing the circumstances which induced the party to call him. Though the answers of the witness may involve him in contradictions calculated to impair his credibility, that is not sufficient reason for excluding the inquiry. * * * Inquiries calculated to elicit the facts, or to show to the witness that he is mistaken and to induce him to correct his evidence, should not be excluded simply because they may result unfavorably to his credibility. In case he should deny having made previous statements inconsistent with his testimony, we do not think it would be proper to allow such statements to be proved by other witnesses."

This court has consistently continued to hold to the rule that a party who calls a witness may not impeach him by calling another witness to prove that on a prior occasion the testifying witness has made contradictory statements or given testimony of a different character. The course pursued on the trial of the case at bar was in direct violation of this rule. Repetto was called for the sole purpose of impeaching the People's witnesses who had failed in the matter of identification, and that was the only effect of his testimony. He testified to no fact that bore directly upon the defendant's guilt or innocence. He simply and flatly contradicted these witnesses, who were testifying that they could not identify the defendant, by stating that in the Coroner's Court they had picked out the defendant as the man who had done the shooting in Belmont avenue and Hughes avenue. If we were permitted to enter the field of conjecture we should say that it is quite as probable that these witnesses were mistaken, or were lying, before the police officers or the coroner, as it is that they testified falsely at the trial. The suggestion that the testimony of Repetto did not tend to impeach these witnesses, but rather to support them, is without merit, since it was no more competent for the one purpose than the other. ( People v. Katz, 209 N.Y. 311; People v. Jung Hing, 212 N.Y. 393.) No rule of evidence rests upon a more generally approved foundation than that it is not permissible for the party calling a witness to fortify his testimony by showing that on other prior occasions he has made similar statements. To this broad rule there are a few recognized exceptions which have no application in the case at bar.

Having demonstrated, as we think, that it was error to receive this impeaching testimony of Repetto, it remains to briefly consider whether it was an error which affected the substantial rights of the defendant. Upon this point there seems to be little room for argument. The one vital, perplexing question in the case is the identity of the slayer of Officer Cotter. As the record stood when the identifying witnesses had completely given their testimony, the identity of the defendant may well have been a matter of serious doubt with a jury of laymen not versed in the technical rules of evidence. The testimony of Repetto may have weighted the scales in favor of conviction. Having in mind this grave possibility we are constrained to hold that justice requires a new trial, and this disposition of the case renders it unnecessary to consider the many other exceptions presented by the defendant.

The judgment of conviction should be reversed and a new trial ordered.

HISCOCK, CHASE, COLLIN and HOGAN, JJ., concur; MILLER and CARDOZO, JJ., dissent.

Judgment of conviction reversed, etc.