dissents and votes to affirm the judgment of conviction, with the following memorandum: When the evidence is viewed in the light most favorable to the People, as it must be (see, People v Contes, 60 NY2d 620; People v Morales, 113 AD2d 956; People v Bigelow, 106 AD2d 448, 449), it is legally sufficient to sustain the convictions, for " 'any rational trier of fact could have found the essential elements of the crime[s] beyond a reasonable doubt’ ” (People v Contes, supra, at p 621, quoting from Jackson v Virginia, 443 US 307, 319, reh denied 444 US 890). Not only was the proof of guilt adduced at trial legally sufficient, but, contrary to the opinion of the majority, I find the evidence was overwhelming.
In sum, the defendant was positively identified at trial by three witnesses, and a fourth said the defendant looked like the perpetrator. Moreover, two of the witnesses who had positively identified the defendant at trial had also separately and conclusively identified the defendant shortly after the crimes were committed. For example, one of the victims assaulted in the bar, Ann Margaret Forde, was being treated for her wounds at Elmhurst Hospital when the defendant was escorted into the emergency room. Upon seeing the defendant, she jumped up and spontaneously identified him as the man who stabbed her, before she observed a police officer walking behind him.
This is not a case where the identification of the defendant as the assailant was dependent upon a split-second observation made by one witness, when there was no apparent reason to take note of the person’s identity (see, People v Caserta, 19 NY2d 18). Although the observations of the assailant by the *401prosecution witnesses, who positively identified him, were made during a stressful episode, their observations were made under adequate lighting conditions, at close proximity and for a sufficient duration of time to belie a claim that all three had mistakenly identified him.
I agree with the majority that a police officer’s bolstering of the pretrial identifications by two of the victims was improper (see, People v Bolden, 58 NY2d 741; People v Trowbridge, 305 NY 471), as was the trial court’s refusal to repeat the curative instruction during the reading back of the officer’s testimony. Nevertheless, the evidence of identification was not only clear and strong, but overwhelming. Consequently, the improper bolstering may be deemed harmless error (see, People v Mobley, 56 NY2d 584, 585). It was also error for the police officer to testify as to conversations with nonwitnesses (see, People v Melendez, 55 NY2d 445; People v Cruz, 100 AD2d 882). However, in view of the overwhelming evidence of guilt, this error does not warrant reversal (see, People v Crimmins, 36 NY2d 230).
Accordingly, the judgment should be affirmed.