Dissenting Opinion by
Spaeth, J.:The difficulty in this case arises from the fact that the judge at the suppression hearing and the judge at the non-jury trial were the same.1
At the suppression hearing the judge, after extended testimony, found that the victim’s identification of appellant at a line-up had been suggested by the circumstances under which the line-up had been conducted. He therefore ordered the identification suppressed. Ordinarily when a hearing judge makes such a finding and order, he goes on to consider whether the victim can identify the defendant for reasons independent of the line-up, as for example because she had a clear look at her attacker and noticed something distinctive about his face or clothing. Depending upon the evidence, the judge then either finds that the victim can make an independent identification of the defendant, in which case he orders that she may do so at the trial, or that she cannot make such an identification, in which case he suppresses any identification testimony by her. In *9the present case, however, the judge did not make such a further finding and order. Instead, after ordering that the victim’s “identification at the time of the line-up be suppressed,” he went on to order: “and further that, at the time of the trial of this case, the victim ... not be allowed to make an in-court identification unless the Commonwealth is able to establish that such identification has not been affected or tainted by the participation of [the victim] in the line-up herein referred to.”
At the trial the judge concluded that the Commonwealth had shown that the victim’s identification of appellant had not been affected by her participation in the line-up. He therefore permitted her to identify appellant as her attacker. In his opinion sur appellant’s post-trial motions, the judge explains the reasons for his conclusion.
First the judge discusses the victim’s testimony that the lighting was sufficient for her to see her attacker, that she saw him for about ten minutes from a distance of no more than two feet, and that during the ten minutes she and her attacker conversed. Then the judge discusses certain discrepancies between the victim’s initial description of her attacker and a later description, stating that he “determined that such factual discrepancies were insignificant.”
So far so good. Then, however, the judge says this:
“Miss Ginyard [the victim] attended a total of three lineups. (T-13). No identification was made at the first two lineups. (T-16). The third lineup was the first in which the defendant was included. The positive identification of the defendant made by Miss Ginyard following the third lineup was positive and without hesitation and remained unshaken during the subsequent hearings and trial. The lineup in which the defendant was identified by the victim occurred within three weeks after the event. Such an interval is not sufficient to shade an otherwise sound identification.”
*10Thus the judge as trial judge referred to the very line-up identification that as suppression judge he had suppressed. Nor was it a casual reference. To the contrary, in characterizing the suppressed identification as “positive,” “without hesitation,” and “unshaken,” he can only be saying that the fact that the line-up identification was so good2 demonstrates, or at least corroborates, his conclusion that the victim had a basis of identification independent of the line-up.
There can be no doubt that if the judge had been sitting with a jury, and the jury had heard the suppressed line-up identification, a new trial would have to be ordered. Commonwealth v. Ehly, 457 Pa. 225, 233-236, 319 A.2d 167, 172-73 (1974). The question, therefore, is whether it matters that the judge was sitting without a jury.
Sometimes it is held that it does matter. Suppose for example the judge hears hearsay testimony. The reasoning is that since he is legally trained, it may be assumed that he knew enough, and was able, to put the testimony out of his mind. This reasoning must always take into account, however, the inherently prejudicial quality of the specific evidence involved, and the importance of that evidence to the particular case. As we said in Commonwealth v. Conti, supra at 501, 345 A.2d at 245,3 collecting and discussing the cases, we do “not simply rely on the fiction of a judicial blindside” but rather look to the facts of each case to decide whether “a trial judge [was] able to maintain his impartiality after hearing incompetent evidence.”
Here the judge’s opinion demonstrates that he was not able to maintain his impartiality. Had he made no reference in his opinion to the line-up identification, the question would be whether as trial judge he could maintain his impartiality, even after he had as suppres*11sion judge heard incompetent evidence. That would be the same question as was presented in Conti and the cases like it, although how it would be answered is by no means clear; there would be good arguments both ways — pro a new trial, that the line-up identification was so prejudicial that we could not rely on judicial blindside, or contra, that it was more in the nature of evidence corroborative of the in-court identification and we could count on the judge putting it out of his mind. Here, however, the judge not only referred to the line-up identification, but, characterizing it in a manner inconsistent with his suppression order, relied upon it.
In these circumstances the pertinent case is Commonwealth v. Oglesby, 438 Pa. 91, 263 A.2d 419 (1970).4 There the trial judge’s statement from the bench disclosed that despite his having found the defendant guilty, he had had a reasonable doubt, which was not dispelled until he examined evidence that would have been inadmissible at the trial, specifically, the defendant’s criminal record. Taking the judge at his word, the Supreme Court said, “[W]e cannot say that the trier of fact was convinced by admissible evidence that [the defendant] was guilty beyond a reasonable doubt.” Id. at 94, 263 A.2d at 420. Accordingly, a new trial was ordered. Here, we cannot be sure that the trial judge was convinced by admissible evidence that the victim could independently identify her attacker. The reason we cannot be sure is that the judge says he arrived at his conclusion partly on the basis of inadmissible evidence. I would take him at his word, and order a new trial.
Hoffman, J., joins in this opinion.. Why defense counsel did not ask at trial that the judge recuse himself is not apparent. See Commonwealth v. Conti, 236 Pa. Superior Ct. 488, 345 A.2d 238, 242-46 (1975); Commonwealth v. Badger, 238 Pa. Superior Ct. 284, 294, 357 A.2d 547, 552 (1975) (Spaeth, J., dissenting).
. One can only wonder, why, then, had he suppressed it?
. Jacobs and Price, JJ., dissented.
. See also Commonwealth v. Molina, 236 Pa. Superior Ct. 598, 346 A.2d 351 (1975) (Spaeth, J., joined by Hoffman and Cercone, JJ. dissenting).