¶1I cannot agree with the court’s conclusion that there was nothing wrong or improper about the pretrial identification procedures employed in this case. On the contrary, I believe that they were highly suggestive and prejudicial, and that they may not be treated as harmless error. The defendants, it is true, were identified at the trial as the robbers — Gonzalez by one witness and Castellano by three—but it is impossible on the record before us to say that these identifications had an independent source and were *59not tainted by the pretrial identification procedures employed by the prosecution.
¶2RE DEFENDANT GONZALEZ
¶3Only one witness, a Mrs. D’Amora, gave identification testimony against Gonzalez. She had been shown his photograph— 16 months after the commission of the crime—just before she entered the courtroom to identify him. She testified that she had observed him but briefly when he confronted her with the gun and that she was extremely frightened at the time. 'Significantly, it was established that when first questioned, immediately after the robbery, she had given the police a wrong description of Gonzalez. She was not asked, and did not say, whether she could have identified him upon the trial had his picture not been shown to her.
¶4RE DEFENDANT CASTELLANO
¶5Upon the trial, Mrs. D’Amora also pointed to Castellano as one of the robbers. Her identification of him is even more suspect than that of Gonzalez. At the time of the robbery, Castellano had a full blown beard, and she had been shown his picture, clean shaven — as he looked at the time of the trial — just before she entered the courtroom. The other two witnesses against him were a Mrs. Pisciotto and a Mr. Bloch; they had viewed Castellano in a showup, as he stood alone in a room at the police station, and were pointedly told by police officers, “ [w]e caught this man, the robber of your store.” Then, as with Mrs. D’Amora, these witnesses, just before they entered the courtroom, were shown Castellano’s picture. One of them, Bloch, frankly acknowledged that he would not have been able to identify Castellano in court were it not for the pretrial procedures.
¶6The pretrial identification procedures thus employed — the showup of Castellano accompanied by the police declaration, ‘ ‘ this was the robber of your store ’ ’, and the exhibition of photographs of both defendants to the witnesses, 16 months after the robbery, just before they went into the courtroom — cannot be justified. They were so highly suggestive and prejudicial as to require that the case be remanded so that the trial court may determine whether the in-court identifications were induced *60or influenced by the pretrial procedures. (See, e.g., People v. Burwell, 26 N Y 2d 331, 336; Foster v. California, 394 U. S. 440, 443; United States v. Wade, 388 U. S. 218.) I note that, although the burden rests with the prosecution to establish, ‘ ‘ by clear and convincing evidence, ’ ’ that the identifications made in court had independent sources (People v. Burwell, 26 N Y 2d 331, 336, supra; see People v. Ballott, 20 N Y 2d 600, 606-607), not the slightest effort was here made to indicate that they were not affected by the tainted pretrial procedures adopted.
¶7Unquestionably, there was in this case more than ample ground for concluding that the in-court identifications were tainted by the “ unnecessarily ” or “ impermissibly suggestive ” pretrial procedures adopted, and that there was ‘ ‘ substantial likelihood of irreparable misidentification. ” (Foster v. California, 394 U. S. 440, 442, supra; Simmons v. United States, 390 U. S. 377, 384; Stovall v. Denno, 388 U. S. 293, 302; see People v. Logan, 25 N Y 2d 184, 191; People v. Rivera, 22 N Y 2d 453, 455; People v. Ballott, 20 N Y 2d 600, 606, supra; People v. Brown, 20 N Y 2d 238, 244; United States ex rel. Phipps v. Follette, 428 F. 2d 912, 914; cf. Clemons v. United States, 408 F. 2d 1230, cert. den. 394 U. S. 964.) No attempt was made to ascertain ‘ ‘ whether, before the imprint arising from the unlawful identification procedure, there was already such a definite image in the witness’ mind that he was able to rely on it at trial without much, if any, assistance from its successor.” (United States ex rel. Phipps v. Follette, 428 F. 2d, at p. 915, supra.) Improper as was the showup of Castellano, even worse was the impropriety involved in the photographic display a year and a quarter after the crime. Most apt is this statement of Judge Friendly, writing for the court, in the Phipps case (428 F. 2d, at p. 915):
“ Lapse of time between the crime and the confrontation is also important; the longer the interval, the greater the dangers that the initial image will have dimmed and that the second image will play a significant role, Also, a long interval between the initial observation and the trial coupled with an improper confronta*61tion a comparatively short time before the witness appears in court enhances the danger that he may be relying on his most recent encounter. ’ ’
¶8In the case before us, there is no warrant for the majority’s conclusory statement that, on the present record, the in-court identifications were based on observations made at the scene of the crime and, indeed, while the witnesses were looking down the barrel of a gun. As in People v. Ballott, so here, their observations took place ‘‘but for a few minutes during a frightening and upsetting episode” (20 N Y 2d, at p. 607). Nor, as I noted above, may the error be stamped as harmless, since without the identification testimony of the three witnesses mentioned, the admissible evidence against the defendants falls far short of being sufficient to support a verdict of guilt.
¶9In short, then, since there is more than enough here to indicate the likelihood that the in-court identifications were improperly based, at least in part, on the pretrial showup and photographic display, the case should be remitted for a hearing to determine whether, in fact, such is the situation. I would, therefore, modify the judgments of conviction to the extent of directing a hearing as to the effect of the pretrial identifications on the identifications made in court.