OPINION OF THE COURT
WEIS, Circuit Judge.In this habeas corpus appeal, a state prisoner contends that a delay of twenty-five months between his arrest and trial violated his constitutional right to a speedy trial. He also asserts that he was denied his right of confrontation when hearsay evidence about lineup identification was introduced by the prosecution. Finding that most of the pretrial delay was attributable to the defense, and that the absence of objection to the identification testimony was a considered, tactical decision by defendant’s counsel, the district court rejected petitioner’s challenges. We agree and affirm.
In January, 1971, a group of eight men, singly and in pairs, entered Dubrow’s Furniture Store in Philadelphia posing as customers. When all were in place, they pulled out guns and announced a robbery. Store employees were bound and several were kicked or pistol whipped. One was doused with gasoline and set afire. Three were shot, one fatally.
Four days later, petitioner Boelter was arrested and identified in a lineup as one of the gunmen by several Dubrow employees. He was tried twice. The first jury was unable to agree upon a verdict, but the second found him guilty of first degree murder. Boelter’s posttrial motions were denied by the Court of Common Pleas and the judgment of sentence was affirmed by the Pennsylvania Supreme Court, Commonwealth v. Boelter, 463 Pa. 162, 344 A.2d 475 (1975).1 Boelter later filed a petition for habeas corpus in the district court. After an evidentiary hearing, the district court, in an extensive opinion, denied the writ. United States ex rel. Boelter v. Cuyler, 486 F.Supp. 1141 (E.D.Pa.1980).2
On appeal the petitioner has raised two issues that were the subject of the evidentiary hearing. They are not related, and we will address each in turn.
*830I
THE SPEEDY TRIAL ISSUE
Boelter was arrested and jailed on January 8, 1971. He was indicted on February 11, and on March 29, the case was listed for trial. During this period, Boelter was represented by Nino Tinari, Esquire. Because he was unable to reach an agreement with the Boelter family on a fee arrangement, the lawyer withdrew from the case on May 24. A few days later, however, the Philadelphia Common Pleas Court designated him as court appointed counsel, despite the fact that he also represented another of the accused, Edward Sistrunk.
Tinari requested and received several continuances during the time from the initial trial setting until May 3, 1972, when he informed the court that Boelter wished to have other counsel. The defendant complained that his lawyer had not consulted with him in 16 months despite numerous written requests. The Common Pleas Court fined Tinari $1,000 for his conduct, or lack of it, in the case,3 and the district court characterized his behavior as “reprehensible.” 486 F.Supp. at 1144 n.5.
During the same period, the prosecutor to whom the case was assigned became ill and the matter was transferred to another assistant district attorney. Further delays resulted from the backlog of older cases and the court’s reluctance to schedule jury trials during holiday periods. After Tinari was discharged, the court appointed Joel Moldovsky, an experienced and capable trial lawyer, to represent the defendant.
Between May and November 9, 1972, when Moldovsky filed a speedy trial motion, the parties obtained four more continuances. In ruling on the motion, the state court denied dismissal but ordered the trial to proceed no later than January 8, 1973. Nevertheless, it was again delayed and on February 20,1973, the defense filed another speedy trial motion. It was denied, but the first trial began on February 26, 1973.
The district court attributed most of the first sixteen months of delay to the defense because of Tinari’s unpreparedness, his fee dispute with the Boelter family, and his representation of the other defendant. The prosecution was charged with only two months and three weeks during the time Tinari was counsel.
The following ten months were divided equally between the Commonwealth and the defendant. Moldovsky required a few months to prepare for trial and moved to have the case transferred to another judge. During this period he also suffered severe illnesses and needed several weeks to recuperate. Five additional months were lost in court backlog and scheduling difficulties but there was no evidence of prosecution efforts to gain a tactical advantage through delay.
Having made these factual findings, the district court analyzed them in light of four factors articulated in Barker v. Wingo, 407 U.S. 514, 530, 92 S.Ct. 2182, 2192, 33 L.Ed.2d 101 (1972): “Length of delay, the reason for the delay, the defendant’s assertion of his right, and prejudice to the defendant.” The district court determined that although the Commonwealth was responsible for some of the postponements,4 the reasons for these continuances were neutral. With respect to the third factor, the court found that Tinari never demanded that the case be tried or the charges dismissed. Only after Moldovsky’s appointment were Boelter’s rights formally asserted.
Finally, the court cited three interests served by the right to a speedy trial that might have been prejudiced by the delay: “(i) to prevent oppressive pretrial incarceration; (ii) to minimize anxiety and concern of the accused; and (iii) to limit the possibility that the defense will be impaired.” 486 F.Supp. at 1151, quoting Barker v. Win-*831go, supra at 532, 92 S.Ct. at 2193. The only asserted detriment to the defendant’s case was the unavailability of an alibi witness, but this man had contradicted Boelter’s statement to the police and then disappeared. Even if found, the witness was unlikely to have been of any help to the defense, and a prompt trial would not have changed that circumstance. We agree with the district court that no prejudice to Boelter’s defense of alibi resulted from the delay and, given the hung jury in the first trial, there was no harm to the defense strategy of asserting misidentification.
The district court carefully weighed and balanced each factor and found the petitioner’s sixth amendment rights had not been infringed. This conclusion was based on a painstaking review of each continuance, the reasons for delay, and possible prejudice to the petitioner. We find no error in the conclusion that the period of time between indictment and trial did not violate the constitutional guarantee of a speedy trial.
II
THE CONFRONTATION ISSUE
The defense in the case was that of mistaken identity. At the first trial, the prosecution produced six witnesses who identified Boelter either at a lineup, in court, or both. In addition, the Commonwealth called Lieutenant Margulis, the police officer who supervised Boelter’s lineup. On cross-examination, defense counsel was able to demonstrate inconsistencies in the descriptions given by the witnesses, as well as equivocal statements made by some of them at the lineups of Boelter and other suspects. Defendant undermined the identification testimony further by recalling several of the prosecution’s witnesses as his own to reveal additional discrepancies. In addition, he called some Dubrow employees who witnessed the robbery-murder but were unable to identify Boelter.
Before any testimony was produced at the second trial, defense counsel, in his opening to the jury, stressed that he anticipated the Commonwealth would call a number of identification witnesses who had been uncertain about Boelter and, in fact, had pointed out others in the lineups. This prediction, however, proved to be faulty because the Commonwealth called only two of the lineup witnesses, DiMeo and Wagenheim, who made in court identifications of the defendant. They were cross-examined on their lineup designations of Boelter and also about their identifications at standups that did not include the defendant. Wagenheim, particularly, was questioned about the fact that he had seen only one robber but had pointed out another man as well as Boelter at lineups.
Lieutenant Margulis then testified that six persons had singled out Boelter at the lineup. In addition to DiMeo and Wagenheim, the officer listed Carson, Block, Sacchetti, and Porecca. The defendant’s lawyer did not object to this hearsay testimony. Before questioning Margulis, defense counsel announced that he wished to reserve part of the inquiry until other witnesses had testified and then recall the Lieutenant as on cross-examination. The court stated: “I don’t think you can do that. You better get your cross-examination in now or you may call him as your witness. That’s the way the law reads.” Transcript at 863. Counsel then sought to question Margulis about identifications by some of the witnesses at other lineups in which suspects other than Boelter were present. The trial judge refused to allow cross-examination of the officer or identification witnesses on this point.
After a lunch recess, the trial judge called a chambers conference and modified his earlier ruling, permitting cross-examination of identification witnesses about lineups in which Boelter did not participate. At this juncture, defense counsel said he would defer further interrogation of Margulis until testimony from all the prosecution’s identification witnesses had been concluded. The trial judge permitted this procedure, thus relaxing his earlier ruling that all cross-examination of Margulis be conducted at one time.
*832After the lawyers returned to the courtroom and the jury was seated, the district attorney, instead of producing further evidence, rested his case and offered the prosecution’s uncalled witnesses to the defense. Defendant’s lawyer did not ask for a reconsideration of the court’s ruling and instead called several witnesses, including Tinari, who had been present at all of the lineups. There was no further cross-examination of Margulis nor were any of the four identification witnesses brought to the stand, even though three had been defense witnesses at the first trial.
The district court held that Boelter had waived his right to assert a confrontation claim by failing to object to Margulis’s hearsay testimony. Had this testimony been excluded, the defendant’s inability to cross-examine Margulis or the declarants about the lineup would not have become an issue. Applying Wainwright v. Sykes, 433 U.S. 72, 97 S.Ct. 2497, 53 L.Ed.2d 594 (1977), the district court concluded that Boelter had failed to show either legal cause for his failure to object, or prejudice from the admission of Margulis’s testimony and exclusion of matters the defense sought to introduce. In these circumstances, defendant’s lack of objection was held to be a tactical decision constituting a “deliberate bypass” of state procedures. Fay v. Noia, 372 U.S. 391, 83 S.Ct. 822, 9 L.Ed.2d 837 (1963).
Although convinced that Boelter had waived the confrontation claim, the district court also held that even if the merits were met, habeas corpus should be denied. The trial judge had assumed that the prosecution would call additional identification witnesses when he ruled on the scope of cross-examination of Margulis and the identification witnesses. Therefore, the district court believed that a request for modification of those rulings after the prosecution unexpectedly rested would have been granted. Finally, even if error had been committed, the court held that Boelter had not been improperly prejudiced by Margulis’s testimony.
On appeal, Boelter asserts that Margulis’s testimony about the identification was proper and, therefore, an objection would have been futile. We turn then to the question of admissibility.
At the time of the trial in 1973, the controlling appellate decision in Pennsylvania was Commonwealth v. Saunders, 386 Pa. 149, 125 A.2d 442 (1956). There, the Pennsylvania Supreme Court found no error in permitting witnesses to testify as to their own previous identifications of the defendant in a lineup. After noting that some jurisdictions admit such evidence, the Pennsylvania court cautioned, “it must be borne in mind that we are dealing here, not with the hearsay testimony of witnesses who merely heard such identifications being made, but with testimony given by those who themselves made the identifications at the lineup.” Id. at 155, 125 A.2d at 445.
Boelter relies upon dictum in Commonwealth v. Hill, 237 Pa.Super. 543, 353 A.2d 870, 877-78 (1975), a decision of the Pennsylvania Superior Court. That case, however, was decided two years after Boelter’s trial, and more than a month after his conviction was affirmed by the Pennsylvania Supreme Court. In Hill, a crime victim stated on cross-examination that he had not been able to identify the defendant. The prosecution then called a policeman who contradicted the victim’s testimony. The trial court sustained a defense objection and instructed the jury to disregard the policeman’s statement. On appeal, the Superior Court found that the curative jury instructions were adequate, but went on to say that identification statements would have been admissible in any event as substantive evidence. Id. at 553, 353 A.2d at 877-78, quoting McCormick on Evidence (2d ed. 1972).5
In later cases the Superior Court has underscored the requirement that the declar*833ant be cross-examined before others are allowed to testify about the identification. In Commonwealth v. Billig, 264 Pa.Super. 199, 399 A.2d 735 (1979), the court found that prior identification hearsay should not have been admitted over objection when the declarants had not been cross-examined. Id. at 203, 399 A.2d at 737. And later in Commonwealth v. Silvers, - Pa.Super. -, 428 A.2d 622, No. 36 Pittsburgh, 1980 (Pa.Super. April 10, 1981), the court determined that it was error to permit a policeman to describe the identifications of three people who had not been cross-examined. In support of its holding that the testimony was objectionable hearsay, the court cited the district court’s opinion in this case. Id. at-, 428 A.2d at 624.
It must be said that at least since 1975, the law in the Superior Court on this evidentiary question has been unsettled. Furthermore, Hill, Billig, and Silvers were not available to defense counsel and to the state court judge at the time of the Boelter trial. The fact that the prior identification exception to the hearsay rule was never cited to the trial court or to the Supreme Court of Pennsylvania on Boelter’s appeal is further indication that Margulis’s testimony was thought to be inadmissible.
It is doubtful that even now the Superior Court would allow identification testimony in a case such as this where the trial judge ruled that the person who made the prior identification could not be called for cross-examination. A holding to that effect would go far beyond the McCormick position cited in Hill, which postulates the availability of the declarant for cross-examination as a predicate for admitting the prior statement. Moreover, the opinions of the Superior Court, an intermediate appellate body, are not binding on the Pennsylvania Supreme Court which has given no indication of departing from its precedents requiring that the witness be the person who made the prior identification.
We are persuaded that a timely objection to the hearsay testimony of Lieutenant Margulis would have been sustained. It is clear, however, that counsel’s silence was not mere oversight. At the hearing in the district court, the petitioner’s trial counsel testified that he was “tickled to death” when Margulis began discussing what some of the identification witnesses did at the lineup since “that was exactly part of the defense.” Having decided neither to object to the hearsay testimony nor to move to strike it, the defense is bound by its own decision on strategy. The Supreme Court has admonished defense lawyers that they may not “take their chances on a verdict of not guilty in a state trial court with the intent to raise their constitutional claims in a federal habeas court if their initial gamble does not pay off.” Wainwright v. Sykes, supra at 89, 97 S.Ct. at 2507.
Even accepting arguendo Boelter’s position that Margulis’s hearsay was admissible, the question then arises whether cross-examination was improperly limited to the events occurring at Boelter’s lineup alone. The trial judge originally would have restricted all cross-examination to that one lineup, but later modified his ruling to permit questioning of identification witnesses about their observations at other sessions. That liberality, however, did not apply to Margulis’s testimony.
It is at least arguable that the trial judge’s ruling on this point was not erroneous. He expressed concern that extensive cross-examination would bring in irrelevant evidence of identification of several other suspects in the crime. Margulis conducted lineups with four of the suspects and an unlimited cross-examination could have included all statements made at each of those lineups.
Generally the scope of cross-examination is a matter entrusted to the trial judge’s discretion and is rarely reversible error. Even if a reviewing court should conclude that inquiry was unduly restricted, that action need not rise to the level of a constitutional deprivation. As we said in United States ex rel. Cannon v. Maroney, 373 F.2d 908, 910 (3d Cir. 1967): “Errors committed during the trial . . . are not subject to review in a habeas proceeding . . . unless ... so conspicuously prejudicial as to *834deprive the defendant of a fair trial.” If the ruling on the scope of Margulis’s cross-examination prejudiced the defendant in any way, it stopped far short of denying him a fair trial.
Boelter argues that the restriction on cross-examination precluded him from revealing misidentifications by the six witnesses Margulis named. But Margulis’s testimony did not present the jury with uncontradicted third party identifications of the defendant. As noted earlier, two of the declarants about whom Margulis testified, DiMeo and Wagenheim, were cross-examined by the defense and their identifications at the various lineups were explored at that time. Additionally, Margulis’s testimony itself revealed all of the weaknesses in the other identifications of which he had personal knowledge. Boelter has never detailed the information that he hoped to elicit from Margulis that would have undermined the other four identifications recounted by the officer. Defendant’s offer of proof did not mention these four, but focused instead on other Dubrow employees who were not called at the second trial.
After the Commonwealth rested its case, defense counsel had to decide whether he should call the witnesses who had singled out Boelter in order to demonstrate that, they had made errors at other lineups in which he was not included. In return for expected favorable responses on that point, however, substantial adverse effects were a certainty.6
Margulis testified that six persons identified Boelter at the lineup — DiMeo, Wagenheim, Carson, Block, Sacchetti, and Porecca. Since DiMeo and Wagenheim testified it was only the other four persons, therefore, who might have been called by the defendant’s lawyer.
At the first trial, Carson had testified for the defense that he selected two participants on the Boelter lineup as “looking good to him” — Boelter and a policeman who had been asked to participate in the stand up. Defense counsel asked no questions about the other lineups but concentrated on Carson’s ambivalence at the Boelter lineup itself. At the second trial, Margulis testified to the same effect, that Carson had indeed chosen the defendant and a police officer from the Boelter lineup. Margulis’s testimony on direct examination, therefore, already revealed the weakness of Carson’s identification.
Block had also been called as a defense witness at the first trial, and testified that she picked the defendant at the lineup because he looked like the robber she had seen. She conceded she was not able to make a positive identification. At the second trial, Margulis accurately recounted Block’s statement, adding that she had mentioned at a different lineup the possible resemblance of another man. Because Margulis’s testimony cast even more doubt on Block’s identification than the defendant had been able to create at the first trial, it is difficult to see that any prejudice resulted from failure to cross-examine her at the second trial.
Margulis testified that Lucille Sacchetti had commented at the lineup that Boelter “resembled” one of the hold up men. At the first trial, she singled out the defendant in the courtroom. She also testified that she had identified Boelter at the lineup, but *835said on cross-examination that it was “possible” that she had used the word “resembles” on that occasion. As recounted by Margulis at the second trial, Sacchetti’s lineup identification lost much of the sting it had at the first trial since Margulis asserted that she had said Boelter only “resembled” the robber. In this instance, then, Margulis’s hearsay testimony in lieu of Sacchetti’s courtroom identification was more of a benefit than a detriment to the defendant.
The only other witness referred to by Margulis was Porecca. At the first trial, he made a positive in court identification of Boelter. The witness said he was struck by a pistol the defendant was holding, and after being doused with gasoline, was set afire by his assailants. Porecca had testified on direct at the first trial that he had not made any other lineup identifications, but did concede on cross that he thought a policeman in a lineup not conducted by Margulis resembled one of the other assailants. Because Margulis had no personal knowledge of this last statement, it was not ' included in his testimony at the second trial. Had Porecca been called to the stand, it was to be expected that he would make an in court identification of the defendant and would repeat the highly adverse details of the crime.
Thus, the jury at the second trial already knew through Margulis’s testimony that the lineup identifications of some of the witnesses were not strong. Had defendant’s counsel called these witnesses as his own, or even as on cross-examination, he knew that some of them would furnish positive courtroom identifications and substantially injure the defendant’s cause. The decision not to call these people as defense witnesses, therefore, is justifiable. There would have been a high price to pay for some additional evidence of mistake in recognizing other suspects, and the gain would have been more than offset by the concomitant damage. As it was, the weakness of the Commonwealth’s evidence had already been put on the record by Margulis and the defense was given the benefit of equivocation by the witnesses without being subjected to courtroom identifications.
Although it is now asserted that these witnesses were not called for the defense because of the Pennsylvania “voucher rule,” we think it more plausible that the decision was taken as a tactical move. The defense did not hesitate to use some of these witnesses at the first trial, despite the same voucher rule. When the prosecution unexpectedly rested at the second trial, the defense had several options. It could proceed as it had at the first trial to call the other witnesses as its own. As an alternative, it could have asked the trial judge to reconsider his ruling in light of the unanticipated shift in the prosecutor’s strategy. As the district court noted, this option was a viable one since the voucher rule is frequently relaxed in Pennsylvania. The defense chose neither of these paths. Instead it decided to capitalize on the situation by arguing to the jury that the reason the prosecution did not call the other employees who had seen the robber was because they could not identify Boelter.
Even if we accept petitioner’s argument that the voucher rule would have been strictly applied, and that it would have been futile to ask the trial judge to reconsider, we agree with the district court that the petitioner was not denied due process. In Chambers v. Mississippi, 410 U.S. 284, 93 S.Ct. 1038, 35 L.Ed.2d 297 (1973), which Boelter cites, application of the hearsay and voucher rules prevented the defendant from introducing highly credible, exculpatory evidence. The harm, if any, suffered by Boelter because of the voucher rule falls far short of that in Chambers. Cf. United States v. Lowell, 649 F.2d 950 (3d Cir. 1981); Government of the Virgin Islands v. Smith, 615 F.2d 964, 970, 972 (3d Cir. 1980).
Our review of the trial record persuades us that if the trial judge erred, that error was harmless. Chapman v. California, 386 U.S. 18, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). The justification for defense counsel’s tactical decision points up the fact that from a practical standpoint, the adverse effect, if any, of the trial judge’s ruling was slight. The defendant was identified in open court by two witnesses who were positive in their *836testimony. Both were subjected to extensive cross-examination. The testimony of the other identification witnesses who were not called would have been cumulative and less favorable to the defendant than Margulis’s hearsay. Any error was therefore harmless beyond a reasonable doubt.
Because we believe the district court properly interpreted Pennsylvania law and properly assessed the effect of the precluded evidence on the outcome of the trial, the judgment will be affirmed.
. In the Pennsylvania Supreme Court, the conviction was affirmed by an equally divided vote. Three of the justices would have remanded for an evidentiary hearing on the delay issue, 463 Pa. 162, 344 A.2d 475 (1975). Whatever deficiencies in that respect may have existed have been cured by the hearing in the district court.
. During the pendency of this appeal, petitioner changed his name to Ahmad Abdus-Sabur. For continuity, throughout this opinion we will use Boelter, the name under which he was tried and appeals were taken to the various courts.
. For some reason not explained by the record, the fíne was later remitted.
. Although it does not affect our ruling on the delay' issue, we note that the Commonwealth was also responsible for part of the delay attributed to Boelter because it appointed Tinari as counsel when he already represented a codefendant. That dual representation was bound to cause delay, if not a conflict of interest.
. McCormick’s treatise states that prior statements of identification are admissible only when the declarant is present in court and “available for cross-examination.” The safeguards that surround staged out of court identifications were cited as an additional reason for accepting evidence of that nature. McCormick on Evidence, supra at 603.
. During his testimony in the district court, Boelter’s trial counsel contended that because one eyewitness said he saw only one of the robbers at the scene but identified the defendant and another person at the lineup, what might have been a misidentification actually was substantive evidence that someone else had committed the crime. Accordingly, the argument went, that evidence should have been admitted. But the record shows that only Wagenheim clearly meets that description and he testified at the second trial. Joseph Ginsberg, another eyewitness who saw only one robber, admitted at the first trial that he picked two suspects at the lineups. But he explained that he had positively identified only Boelter. His description of the second suspect was that he “looked like the man [Boelter] that I had picked in the first line-up.” Transcript at 273. Even if Ginsberg’s statements are considered inconsistent, their exculpatory effect would in no way have approached that of the excluded statements in Chambers v. Mississippi, 410 U.S. 284, 93 S.Ct. 1038, 35 L.Ed.2d 297 (1973). Consequently, we find no merit to this contention.