In 1953, a jury in Florida state court found Jimmy Ray Bonds guilty of rape. He was sentenced to thirty years imprisonment, which he did not appeal. In 1974, however, Bonds challenged his conviction by seeking a writ of habeas corpus, first, unsuccessfully from Florida, and then in the federal courts.1 The district court denied relief. In his appeal, Bonds challenges the lower court’s decision in only one respect: he asserts that his trial counsel failed to render him constitutionally effective assistance in connection with the decision whether or not to pursue a direct appeal of his conviction.
The State of Florida was unable to provide any transcript of Bonds’ trial or sentencing. In any case, a transcript would not be likely to reveal whether Bonds’ counsel did or did not adequately advise and consult with him about the possibility of appeal. The lower court, however, conducted an evidentiary hearing, at which Bonds and his trial attorneys, Murray Klein and Harry Durant, testified. The findings of the lower court bespeak its effort to identify the facts of a case which did not readily fit within more common categories of effective or ineffective assistance. We therefore will set out the relevant findings at some length.
The lower court found that Bonds “did not affirmatively consent to forego an appeal, nor did he press for an appeal.” Rather, Bonds, then sixteen years of age, “left that decision to Klein;” he “reposed his confidence in his lawyer to make the right decision.” “Klein, in consultation with Durant, exercised that judgment for” him. Though the opinion below does not expressly so find, its conclusion that Bonds relied on his lawyer may well imply that Bonds was at least aware that some sort of appeal option existed. At one point, the court indicated that “[tjhere is a strong suggestion . . . that the motivating factor which caused Bonds not to take an appeal was the fact that Bonds had received a thirty year sentence in a capital case where the facts were harsh facts indeed.” By this the court may have meant that Bonds’ reliance on his lawyers seemed to be accompanied by, or to stem from, his awareness of his jeopardy. But the court did not find this as a fact. (It is also possible that the court meant only that Bonds’ counsel took into account the risks he ran if he appealed, rather than that Bonds himself did.) In some respects, at least, the court did believe that Bonds’ understanding of his plight was limited for it twice commented that Bonds “was not familiar with his appellate rights.” Bonds had testified that he “rememberfed] *1128shortly after the conviction Mr. Klein saying something to the effect he thought there were errors which would warrant a new trial, and I think he was even talking to Mr. Durant other [sic] than me . . .” The court concluded that Bonds’ lawyers had, in his presence, given some consideration to the possibility of appeal, but it also specifically referred to his lack of familiarity with his rights at that time. The court also observed that “there was at least some discussion relating to the possibility of appeal.”
Before we can evaluate the merits of Bonds’ position, we must consider whether the lower court’s view of the facts is correct. Although neither party to this appeal explicitly attacks the findings below, the state’s brief outlines a seemingly different picture of the events of the case. It argues that appellant failed to sustain his burden of proof, and that “the evidence strongly supports . . . conclusions (1) that the appellant was apprised of the right to appeal and (2) that the appellant agreed to forego that appeal in view of the risks involved.” To be sure, the state does not assert that Bonds’ lawyers acquainted him with the procedural requisites for prosecuting an appeal; in this sense, the state does not depart from the court’s view that Bonds “was not familiar with his appellate rights.” But the state argues that Bonds evaluated the risks and as a result agreed not to pursue an appeal. The court may have believed that Bonds was cognizant of the risks, and that he was aware in some sense of the possibility of appeal. Even if Bonds took these factors into account in choosing to rely on his lawyers’ judgment, the court clearly found that Bonds “did not affirmatively consent to forego an appeal” but rather relied on his lawyers, who “exercised that judgment for him.” (Emphasis ours.)
We, of course, are bound by the lower court’s findings of fact unless these are clearly erroneous. These are not. Only Bonds claimed to have any distinct memory of the events in question. He testified that he was never advised of his appellate rights, and understood nothing about the taking of an appeal, and that he felt his lawyer would have pursued an appeal if the lawyer had felt it would help. Moreover, the thrust of Bonds’ testimony was contradicted by both of his lawyers. They testified that as a matter of custom they would have discussed the appeal decision with their client, though neither had any distinct memory of whether or not they had done so. The state also mounted a direct attack on Bonds’ credibility. Emphasizing Bonds’ long delay before seeking relief, the state suggested that the impulse for his present effort was not new knowledge of his legal rights, but hope that the transcript of his trial would have disappeared — as it had — and that relief would therefore be more likely. While the state’s portrayal of Bonds is not implausible, it is also not entirely compelling. Judge Eaton, of course, heard Bonds’ testimony, as we did not. Perhaps guided by his own knowledge of 1950s Florida criminal procedure, to which he referred briefly at the hearing.2 the lower court did not have to believe Bonds’ somewhat inconsistent testimony entirely (or at all) — but it could reasonably have accepted that at least a central point was true — that Bonds had not actually taken part in the decision on whether or not to take an appeal.
Taking these facts as established, we must now decide whether they show a violation of Bonds’ right to counsel. Our starting point is the observation that no such violation can have occurred unless Bonds had a right to effective assistance of counsel on appeal in 1958, when these events transpired. It is established that the indigent defendant’s right to appointed counsel at trial, and on appeal is retroactive, See, e.g., Stovall v. Denno, 388 U.S. 293, 297-98, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967); Smith v. Crouse, 378 U.S. 584, 84 *1129S.Ct. 1929, 12 L.Ed.2d 1039 (1964) (per curiam). Conceding as much, one might still argue that the particular elements which this Circuit has come to view as essential to effective assistance on appeal should not carry retroactive force. Instead, once a court ascertained that a defendant had had counsel physically present, it would evaluate counsel’s effectiveness according to the norms of representation prevailing at the time of counsel’s actions. Cf. United States ex rel. Williams v. Pennsylvania, 378 F.Supp. 1295, 1297 (E.D.Pa.1974). See generally Burston v. Caldwell, 506 F.2d 24, 27 (5th Cir.), cert. denied 421 U.S. 990, 95 S.Ct. 1995, 44 L.Ed.2d 480 (1975). We need not decide, however, whether our modern descriptions of the nature of effective assistance diverge significantly from those which governed in 1953, for a panel of this Court has already declared in broad terms that the present criteria for effective assistance do apply retroactively, Bailey v. Ault, 490 F.2d 71, 72-73 (5th Cir. 1974) (per curiam); see Byrd v. Smith, 407 F.2d 363, 365 n. 2 (5th Cir. 1969). Compare Shiflett v. Virginia, 447 F.2d 50 (4th Cir. 1971) (en banc), cert. denied 405 U.S. 994, 92 S.Ct. 1267, 31 L.Ed.2d 462 (1972) (rejecting retroactive application of duty of attorneys to advise defendants of their appellate rights).
Even when a rule of law has retroactive effect, however, its application may not be automatic. The court’s capacity to evaluate past conduct in terms of present rules is stretched to the utmost when, as here, a transcript of the original events can be no longer provided and the participants’ memories have also faded. In one of its first efforts to cushion the retroactive impact of newly-declared constitutional protections, the Supreme Court put considerable emphasis on the unavailability — through no fault of the state — of a trial transcript. The case was Norvell v. Illinois, 373 U.S. 420, 83 S.Ct. 1366, 10 L.Ed.2d 456 (1963). The issue arose out of the response of the State of Illinois to the Supreme Court’s decision in Griffin v. Illinois, 351 U.S. 12, 76 S.Ct. 585, 100 L.Ed. 891 (1956) which reasoned that the state must provide to an indigent defendant a right of appeal comparable to that enjoyed by the more wealthy, and therefore proscribed the flat denial of a free transcript if that denial stymied an indigent’s appeal. Illinois had thereupon provided by rule that indigent defendants, convicted both before and after Griffin, could receive a free trial transcript, unless the transcript could no longer be produced. In Norvell’s case, no transcript of his 1941 trial existed, and an attempt to reconstruct the record through the testimony of persons who attended the trial was largely unavailing. The Illinois courts declined to afford him a new trial.
The Supreme Court affirmed. Writing for the Court, Justice Douglas emphasized that Norvell was represented by counsel at trial and presumably had his continuing services for purposes of appeal, yet failed to pursue an appeal. In the face of the impossibility of ascertaining what errors, if any, were made at Norvell’s trial, the Court decided to sanction a presumption that his lawyer could have protected his rights on appeal, if (the court implied) legal (and tactical) reasons for an appeal had existed.
The lower court felt that the rationale of Norvell was applicable to Bonds’ case. Indeed, the court indicated that Bonds’ position was less compelling than Norvell’s. The Supreme Court had had to rely on a presumption that Norvell had available the services of a lawyer for appeal. No such presumption, the court stated, was needed as to Bonds, for “[w]e know from the testimony that Bonds had a lawyer and that the lawyer did not refuse to represent Bonds on appeal.” The force of this distinction is questionable. If, indeed, Bonds was denied the effective assistance of counsel on appeal, then his ability to exercise his appellate rights could have been jeopardized. It is true that the Norvell court itself identified only a refusal by trial counsel to take an indigent’s appeal as posing a “different case.” 373 U.S. at 422-23, 83 S.Ct. 1366. But the presumption of the availability of counsel’s services seems effectively undermined by constitutionally inadequate (albeit available) service as well. Cf. Stokes v. Peyton, 437 F.2d 131, 135-36 (4th Cir. 1970).
*1130A more fundamental objection to application of Norvell to the present case is that Norvell represents an ad hoc approach to problems of retroactivity, which the Supreme Court has not ordinarily utilized since. In effect, Norvell decides that Griffin v. Illinois is not retroactive if the state is, without fault, unable to provide a transcript of the trial. This, loosely, is an example where complete retroactivity would have a burdensome impact on the administration of justice. Such burdensome impact is obviously cause for concern and, in subsequent retroactivity cases, the Supreme Court has viewed this danger as one of the critical factors for decision. See Adams v. Illinois, 405 U.S. 278, 284, 92 S.Ct. 916, 31 L.Ed.2d 202 (1972); Stovall v. Denno, 388 U.S. 293, 297, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967). But these cases generally evaluate the impact on the administration of justice, as well as other factors, as they will be felt over the full range of potential litigation. The resultant decision holds a rule retroactive or prospective, but not “both at different times.” Moreover, a major theme of the decisions which have given rules retroactive effect has been the impact of the rule in question on the integrity of the factfinding process. See Williams v. United States, 401 U.S. 646, 653, 91 S.Ct. 1148, 28 L.Ed.2d 388 (1971). The Williams court comments that a “severe impact on the administration of justice has [not] sufficed to require prospective application” in such instances. Id. Though less directly related to the assurance of accuracy than the presence of a lawyer at trial, the assistance of counsel on appeal has been seen as implicating the same concerns. See Stovall v. Denno, supra, 388 U.S. at 297-98, 87 S.Ct. 1967. Hence we conclude that the absence of a transcript of Bonds’ trial does not prevent us from addressing his claims. We are confirmed in this conclusion by the past cases which have, admittedly without addressing Norvell, entertained petitions for relief despite the unavailability of a transcript. See Edge v. Wainwright, 347 F.2d 190 (5th Cir. 1965); cf. Collier v. Estelle, 488 F.2d 929, 933 (5th Cir. 1974).
Bonds’ lawyers were not originally appointed by the state. We measure their conduct as if they were court-appointed, however, because after trial they sought, and received, court appointment and payment from the state. See Goforth v. Dutton, 409 F.2d 651, 653 (5th Cir. 1969). We turn, therefore, to the several cases in this Circuit which have followed the proposition that a court-appointed lawyer must inform his client of his appellate rights. See Daniels v. Alabama, 487 F.2d 887 (5th Cir. 1973), (per curiam); Lumpkin v. Smith, 439 F.2d 1084 (5th Cir. 1971). Cf. Thomas v. Beto, 423 F.2d 642 (5th Cir. 1970) (per curiam). But cf. Collier v. Estelle, 488 F.2d 929, 932 (5th Cir. 1974) (testing waiver of appeal by standards for waiver of counsel, the court noted the absence of indication that the defendant “suggested, acquiesced in, or concurred with decision to dismiss appeal.”) See also Giles v. Beto, 437 F.2d 192 (5th Cir. 1971) (per curiam) (guilty plea). Perhaps the most apposite precedent is an earlier case, Wainwright v. Simpson, 360 F.2d 307 (5th Cir. 1966). There the defendant’s lawyer, perhaps because of the threat of a death penalty on retrial, failed both to prosecute an appeal and to advise his client of the grounds for appeal and the relevant procedural time limits. The court concluded that, however laudable his motive, counsel had no right to deliberately forego his client’s post-trial remedies without consulting with him or obtaining his consent.
The court below recognized, though it did not entirely spell out, the relevance of Wainwright v. Simpson to Bonds’ claim, and indeed implied that, were it not distinguishable (on grounds we shall consider infra ), Simpson would be dispositive. Bonds’ counsel, like Simpson’s decided not to appeal without obtaining the client’s consent. The lower court did not make a finding that Bonds’ counsel had consulted with him either (though the court may well have felt that Bonds had some awareness of the nature of the choice he faced). One basis on which the lower court distinguished Simpson was its finding that “there is evidence . that some consultation was had regarding appeal in the presence of Bonds.” *1131The court does not state, but perhaps we may speculate, that this consultation could have been in part directed to Bonds’ ears or open to his participation. Wainwright v. Simpson does not define what degree of consultation with his client counsel must engage in, but we recognize that some minimum level of consultation must be afforded.
Here, the district court concluded that Bonds was not familiar with his appellate rights, and had left the appeal decision to Klein, in whom he was reposing his confidence. The court described the decision as being made by “Klein, in consultation with Durant.” Bonds “did not affirmatively consent to forego an appeal, nor did he press for an appeal.” We take the implication of these findings to be that little, if any, genuine consultation took place in Bonds’ case. Moreover, Bonds’ lack of familiarity with his appellate rights — though the court does not spell out the extent of this ignorance — may well reflect a violation of the rule of such cases as Daniels, supra, and Lumpkin, supra, that the lawyer must inform his client of these rights. We therefore hold that Bonds did not receive the effective assistance of counsel.
The lower court also sought to distinguish Wainwright v. Simpson on the ground that Simpson’s counsel believed meritorious grounds for an appeal were present. In Bonds’ case, it declared, “[t]here is no showing . . . that Bonds’ court-appointed counsel considered that meritorious grounds were present.” Bonds did testify that one of his lawyers had said “something to the effect he thought there were errors which would warrant a new trial.” Bonds himself remembered the introduction, over objection, of unsigned confessions. One of Bonds’ attorneys remembered that he felt it unfair that Bonds was convicted while the same trial ended in mistrial for a codefendant. The court did appear to accord some weight to Bonds’ memory of his attorney’s comment. Nevertheless, the court was not bound to conclude that Bonds had established his attorneys’ belief that there were meritorious grounds for an appeal.
What is more debatable is the legal significance of this finding. Wainwright v. Simpson, after all, refused to approve an attorney’s unilateral decision that an appeal was not in his client’s best interests. The thrust of this decision ought equally to reach unilateral decisions that no meritorious ground for appeal exists. See generally Anders v. California, 386 U.S. 738, 87 S.Ct. 1396, 18 L.Ed.2d 493 (1967). Moreover, this Court has declined to require that a state prisoner, seeking habeas corpus relief because of a denial of effective counsel, make a showing of some chance of success on appeal. Lumpkin v. Smith, 439 F.2d 1084, 1085, (5th Cir. 1971). Cf. Rodriquez v. United States, 395 U.S. 327, 329-30, 89 S.Ct. 1715, 23 L.Ed.2d 340 (1969); Nelson v. Peyton, 415 F.2d 1154, 1159 (4th Cir. 1969), cert. denied, 397 U.S. 1007, 90 S.Ct. 1235, 25 L.Ed.2d 420 (1970). But cf., e.g., Hooks v. Roberts, 480 F.2d 1196 (5th Cir. 1973), cert. denied, 414 U.S. 1163, 94 S.Ct. 926, 39 L.Ed.2d 116 (1974) (claim of failure of counsel to raise various points on appeal denied since points without sufficient merit to warrant pressing). Though analytically distinct, a requirement that petitioners show that their counsel thought there were meritorious grounds would reflect less solicitude for such claims than the result in Lumpkin manifests.
The lower court, however, felt that another difference between Bonds’ case and Simpson’s was the most important. The court emphasized that Bonds had “left the decision of whether or not to appeal to his lawyer . . .” That Bonds relied on his lawyer might mean that he waived his right to advice and consultation more ample than what he received. It is clear that the right to counsel can be waived altogether. Presumably the right to effective assistance of counsel can also ordinarily be waived. While we might hesitate to give effect to a waiver that purported to permit counsel to fail to prepare a defense, a waiver of effective assistance on appeal is far less drastic. Bonds’ reliance on his attorneys did not free them of the duty to diligently protect him, but only increased *1132their responsibility. It seems reasonable to permit a defendant to decide that he would play no fruitful role in a decision whether or not to appeal. Cf. Bennett v. Mississippi, 523 F.2d 802 (5th Cir. 1975) (implicitly recognizing possibility of waiver of right to counsel on appeal); United States v. Garcia, 517 F.2d 272 (5th Cir. 1975) (waiver of right to conflict-free representation permitted).
But such waiver should be no more readily found by the courts than any other waiver of the right to counsel. Hence waiver must be an intelligent, understanding, and voluntary decision. A valid waiver of the right to effective assistance of counsel therefore would have to be preceded by an explanation to the client of what he was waiving. This explanation should make clear to the client the significance of what he is waiving and the risks he runs. See generally United States v. Garcia, 517 F.2d 272, 277-78 (5th Cir. 1975) (directing trial court to “carefully evaluate” waiver of right to conflict-free representation, by using procedure akin to that provided in guilty plea proceedings by F.R.Cr.P. 11.)
Although the district court spoke of reliance rather than waiver, its evaluation of Bonds’ claims throws considerable light on the waiver question as we have formulated it. The testimony of Bonds’ lawyers was that they could not remember how they had handled Bonds’ case but that by custom they would have put the appeal decision to him. Only Bonds’ testimony — supplemented, perhaps, by Judge Eaton’s own awareness of standard practices of that day — supported his claim of ineffective assistance. Thus the court’s response may well reflect an evaluation of Bonds’ veracity. That response did nob go as far as to find a violation of Wainwright v. Simpson, but many of the court’s findings pointed in that direction. Even the court’s conclusion that Bonds had relied on his lawyers did not require total disbelief of Bonds, for at one point on the stand, he had said as much.
The court made no findings concerning the extent to which Bonds had or had not been so carefully advised of his right to effective assistance that he could validly waive it. Bonds himself, however, testified that he “didn’t know anything about an appeal,” that his lawyer “didn’t mention anything about an appeal at any time.” If this testimony is credible, then it would be difficult indeed to find a valid waiver.3 We understand the district court to have concluded that Bonds’ testimony contained at least some truth. We also understand its sense of the events surrounding the appeal decision as this: that Bonds’ attorneys were undoubtedly solicitous of Bonds’ interests, but were acting relatively independently. We recall as well, the court’s intimation that lawyers of that day tended to make the appeal decision unilaterally. We conclude that the likelihood is that in 1953 even these attorneys, solicitous of their youthful, no doubt distraught, client’s interests, did not counsel him in a way that can now support a finding of a knowing waiver.
Our decision in this case has been closely guided by the district court’s delicate evaluation of these long-forgotten facts. Nevertheless, we do differ with the district court on the law, and conclude that Bonds has demonstrated that he received constitutionally ineffective assistance of counsel. He is therefore entitled to release, unless the state chooses to retry him within a time certain.
REVERSED.
. The state recognized, in a memorandum of law filed below, that Bonds’ state habeas petition raised substantially the same grounds as his federal court petition. In this court, similarly, the state at no time urges that exhaustion considerations mandate a remand of the entire case. The state does object on exhaustion grounds to Bonds’ reliance on Anders v. California, 386 U.S. 738, 87 S.Ct. 1396, 18 L.Ed.2d 493 (1967), but Bonds did cite the case in the state petition, and for substantially the same point. Other exhaustion issues have been raised by the state.
. He commented then that, “This idea of explaining appellate rights at the time of sentencing has come into the law long after this situation . .. The lawyer was just more or less to decide whether or not he wanted to take an appeal . . . and so courts did not go around seeing if somebody had all his appellate rights.” (R. 54-55).
. The evidentiary hearing did not bring out whether Bonds might at some point have told his attorneys in broad terms that he could not take any meaningful role in the legal decisions to be made. Even if he did, however, we would be hesitant to accept the effectiveness of such a general waiver, forswearing a role in a decision whose parameters were as yet invisible. See Bennett v. Mississippi, 523 F.2d 802, 804 (5th Cir. 1975).