concurring in part and dissenting in part:
I concur in Part IV of the majority opinion which affirms Hayes’ conviction. I dissent from Part II, which rejects Cecil’s claim of prejudicial error in the district court’s refusal to allow cross-examination of Creswell respecting his aborted insanity plea; and from Part III, which rejects the several defendants’ claim of error in the district court’s summary denial of their fair cross-section challenge to the jury array.
I take the refusal to allow cross-examination and jury challenge issues in that order.
I
It is hard to imagine evidence — other than directly exculpating evidence — more critical to a criminal defendant’s defense than evidence that a key witness against him is not credible on the matters in issue because of a mental impairment affecting his powers of recall or general reliability. For this reason, despite the obvious added risks of embarrassment and distress to the witness, the rule is that under appropriate circumstances such evidence may be admissible, either by cross-examination of the witness or by independent proof or both. See United States v. Lopez, 611 F.2d 44 (4th Cir.1979). Whether, assuming relevance, mental impairment evidence should be admissible in a particular case, given its special risk of prejudice to a witness, is of course committed in the first instance to the discretion of the trial judge, operating under the general guidance of Fed.R.Evid. 403, of which the Lopez rule is simply a special application. Under Rule 403, such evidence, like any, “although relevant, may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by consideration of undue delay, waste of time, or needless presentation of cumulative evidence.” Lopez, emphasizing the special added danger of unwarranted prejudice to the witness posed by this type evidence, requires that in exercising discretion in this context, account be taken of whether the asserted *1457impairment existed during the period in issue, whether the impairment went to the witness’ competency and ability to recall, and whether its revelation would introduce a confusing collateral issue that would require allowing the Government to introduce explanatory evidence. Id. at 46. And as a further safeguard, Lopez requires that a party seeking to introduce such evidence by any means, make a preliminary offer of proof of the mental impairment evidence, in order to lay a threshold basis for the court’s discretionary ruling. Id. Obviously, a ruling against admission is then reviewable only for abuse of discretion, with all that that implies for great deference to the trial court’s special role and competence in such matters. Id.
So, we start with the proposition, which I readily accept, that solicitude for privacy interests and for trial efficiencies give trial courts great latitude for excluding this type evidence. Nevertheless, I am persuaded that under the special circumstances of this case, discretion was abused here; that Cecil should have been allowed to attempt the limited impeachment by cross-examination that he ultimately sought to make, and that disallowing it denied him the fair trial to which he was entitled.
To start with, any evidence significantly probative of this particular witness’ lack of credibility was absolutely critical here. As the majority opinion concedes, not only was Creswell a key witness against Cecil, he was the only witness whose testimony directly linked Cecil to the conspiracy; there was little, if any, circumstantial evidence. Effectively, Cecil’s defense was that Cres-well’s testimony, the only significant evidence against him, was too unreliable to persuade a reasonable fact-finder of guilt beyond a reasonable doubt. And, as bearing directly and heavily upon this, Cecil had indisputable evidence, properly proffered to the court, that Creswell had within the past two years asserted, officially and publicly, that within three years of the events to which he testified he had indeed been mentally impaired — so impaired that he himself should be relieved of criminal liability in another matter. The district court’s ruling, or series of rulings, effectively denied Cecil the opportunity to lay any of this critical impeachment evidence before the jury.
The district court’s discretion was wide here, but with liberty at stake on the credibility of this single, indisputably criminal, obviously erratic and self-interested witness, I would hold that it was abused here in not allowing Cecil the limited proof that he ultimately sought to make on the matter.
Because trial court discretion depends upon circumstance, it is important in reviewing its exercise to get the context right. And on this, I simply disagree in critical respects with the majority’s reading of the record.
On the majority’s view, the ruling under review was precipitated in mid-trial by a belated proffer of proof by Cecil which could have been made much earlier. It is then asserted that what Cecil wanted at that point was to introduce the psychiatrist’s contemporaneous certificate and his 1981 letter reports filed in the aborted New York insanity plea proceeding as direct evidence of Creswell’s present mental impairment, and to offer the psychiatrist as a live witness for the same purpose. The majority then goes to great pains to demonstrate the inadmissibility of the current “certificate” and the 1981 reports; to question whether the psychiatrist was in fact available as a live witness; and to deride — apparently on medical grounds — the probative value of the psychiatrist’s opinions, as reflected in the documents, of the narcissistic nature and Oedipus-complex origins of Creswell’s impairment. On that basis, the majority then concludes that the only evidence actually excluded by the district court was the record of the Rule 12.2 proceeding in the New York prosecution, which included the psychiatrist’s contemporary reports, and the psychiatrist’s current “Certification”; that the district court was never called upon to make any wider ruling excluding “all testimony, cross-examination and argument” related to Creswell’s mental condition, as Cecil asserted on appeal; and that because neither of the only two proffered items of documentary evidence was admissible, there was no error in the *1458limited ruling which only excluded them. Majority op. 1438, 1444.
With all respect, this misreads the record in ways critical to a proper assessment of the district court’s discretionary ruling.
In the first place, the ruling was not precipitated by a belated effort by Cecil to get the evidence in, but by the government to keep it out. The critical information, the linchpin evidence, was that Creswell had in fact given official notice in 1981 of his intention to enter an insanity plea and had then undergone psychiatric evaluation which tentatively supported his plea. This fact was discovered by defense counsel and communicated to the government, which claims not then to have known of it, well in advance of the government’s use of Creswell as a witness. See J.A. 231, 232, 251.14 Cecil’s counsel obviously was under no obligation to do more than tell the government, “we have this information, here is the proof in documentary form; be suitably advised,” by way of laying a basis for making whatever use of it might be appropriate and permitted at trial. It defies belief, and the government does not represent, that it only became aware at midpoint of trial that Creswell might be cross-examined on the point. Yet its motion in limine to suppress the evidence was only brought on for hearing just before (or possibly after) it put Creswell on the stand. See J.A. 217, 218. To the extent that discretion to exclude the evidence might properly take into account the lateness of the issue’s raising by defendant — as the majority suggests — any tardiness here simply could not be laid to Cecil.
Much more critically, the majority opinion, with all respect, is simply wrong in its assertion that the district court's exclusionary ruling was effectively limited to the two documentary items incorporating psychiatric evaluations and that this was the limit of the defendant’s proffer of proof. In fact, by a series of rulings the court did — as Cecil rightly asserts — ultimately exclude any and all evidence, whether documentary or testimonial, whether to be adduced on cross or direct examination, having to do in any way and to any extent with the fact that Creswell had given notice of intention to file an insanity plea. As indicated, the process of exclusion was an evolving one over the two day period devoted to the government’s motion, but its ultimate scope and its effect was unmistakably total exclusion. Most critically, it excluded all opportunity to conduct any cross-examination on the point, which was the only means for developing it actually sought from start to end by Cecil.
The government’s formal motion in li-mine was directly addressed only to the anticipation that defense counsel would cross-examine Creswell about the 1981 insanity plea notice and the related psychiatric evaluation. J.A. 207. It was only to the possibility of cross-examination based upon defense counsel’s known knowledge of the indisputable facts of that proceeding that the government sought an exclusionary ruling. J.A. 209. The thrust of the government’s initial oral argument in support of its motion was only to forestall cross-examination based upon defense counsel’s knowledge. J.A. 220.
At this point, defense counsel had made no proffer of evidence pursuant to the Lopez requirement. It was only to meet that requirement that Cecil first proffered the two documentary items — as substantiation for the propriety of allowing this potential*1459ly embarrassing and possibly collateral matter to be explored on cross-examination. Cecil obviously would have seized at that point upon any opportunity given him to supplement his cross-examination by introducing on direct either the documentary evidence or the psychiatrist’s oral testimony. But his opposition to the government’s motion in limine specifically sought only to protect his right to cross-examine on the basis of the information known both to him, and through him, to the government, and he never at that point sought a ruling admitting the documentary evidence, though he offered to put the psychiatrist on for voir dire if the court desired. J.A. 266, 278.15 In fact, as consideration of the motion continued, Cecil emphasized again and again the narrow scope of cross-examination he proposed and would bind himself in advance to conduct. He would limit it to two to three minutes duration and would seek by it basically only to bring out that Creswell had indeed given notice of an insanity plea in the New York proceeding, and then to pursue the inevitable admission only in severely limited ways. He specifically proposed to explore only the extent to which Creswell had participated with his counsel in the decision to give notice of the insanity plea, to explore Creswell’s recollection of the nature of the psychiatric evaluations then made, to inquire whether Cres-well considered himself to have been mentally impaired at that time and subsequently. See J.A. 235, 266, 267, 285.16 He specifically disavowed any purpose to introduce in evidence either the psychiatric reports or the psychiatrist’s oral testimony. J.A. 278. When the court expressed concern about opening the door to comparable cross-examination by Cecil’s co-defendants, counsel for those defendants expressly waived any right to do so. J.A. 287.
Cecil’s counsel also expressly identified and committed himself to limit the arguments he would make based upon any evidence developed by this limited cross-examination. He proposed to exploit it only by arguing alternatively that Creswell was either honestly doubtful about his mental condition in the relatively recent past, or was prepared then deliberately to misrepresent his condition in order to manipulate the judicial process. J.A. 224-25. He expressly committed not to argue that Cres-well was insane — at any time. J.A. 287. In this light, it is significant that Cecil’s proffer identified a quite legitimate impeachment purpose as an alternative to mental impairment. The district court first tentatively, J.A. 260, then finally, J.A. 290-91, refused to allow cross-examination touching on the insanity plea for any purpose, no matter how limited.17
*1460It is obvious that impeachment by cross-examination on either of these bases could have been highly effective here. Contrary to the majority’s apparent fail-back suggestion of harmlessness of any error in its exclusion, Majority op. 1442-43, it might obviously have tipped the balance of doubt about the credibility of a witness so manifestly disposed as Creswell to criminal conduct and to protect his own interests at the expense of others caught up in criminal charges. The appropriate test for harmlessness of error in the exclusion of this evidence is whether we can say “with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not swayed by the error.” Kotteakos v. United States, 328 U.S. 750, 765, 66 S.Ct. 1239, 1248, 90 L.Ed. 1557 (1946). I cannot say this with fair assurance here. The government's case against Cecil was utterly dependent upon Creswell’s otherwise shaky credibility. Cecil was entitled to impeach it by any effective means that could be kept in reasonable bounds as this clearly could have been.
I believe that the district court unfortunately failed adequately to consider the narrowly limited range of cross-examination and argument actually proposed here; instead became unduly concerned with the collateral consequences of admitting evidence that the defendant did not propose to offer when the motion in limine was being considered; and in consequence deprived the defendant of a legitimate opportunity to impeach the sole witness against him by cross-examination on a point that might well have swayed the jury's judgment. I would therefore vacate Cecil’s conviction and remand for a new trial.
II
Without minimizing the seriousness to Cecil of the district court’s discretionary refusal to allow him possibly vital cross-examination, the district court’s summary dismissal of the defendants’ challenge to the jury selection process presents the more important issue for precedential purposes. For that ruling, as affirmed by the majority here, has critical implications both for the ongoing administration of the Jury Selection and Service Act of 1968, 28 U.S.C. § 1861, et seq. (the Act), and for future applications of the underlying “fair-cross-section” guarantee of the sixth amendment that the Act implements. Because I disagree with the majority’s interpretation of both the Act and the underlying constitutional right in critical respects, I dissent from its affirmance of the district court’s ruling and from its reasoning to that result.
In order to explain my disagreement, it is first necessary to be clear about the exact basis of the district court’s ruling and the majority’s alternative bases for affirmance —one “of law,” essentially that relied upon by the district court, and the other a factual basis not addressed at all by the district court.
In the district court the defendants properly moved under § 1867(a) to dismiss the indictment for “substantial failure to comply with the provisions of [the Act] in selecting the ... jury.” In support of this motion and in compliance with § 1867(d), defendants filed a sworn statement of facts which, they asserted, “if true, would constitute” such a “substantial failure.” Specifically, it was asserted as fact that the juror lists used under the district court’s formal Plan were drawn exclusively from Maryland voter registration lists; that recent voter registration lists contained only 60% of those qualified by age to vote, hence to serve on juries; that this therefore automatically excluded 40% of the prospective jurors within the district; that among other groups, blacks and other racial minorities and persons aged 18-34 tended to register as voters in lower percentages (not specifically known) than did the voting-age population at large; that therefore the 40% of those excluded from jury service by exclusive reliance on voter registration lists probably contained disproportionately large numbers of members of these groups; and that the overall consequence of these facts was believed to be a significant underrepre-sentation of these particular groups on juror lists, thereby violating the fair-cross-*1461section requirement of the Act and the sixth amendment to the Constitution.
In an effort to develop more specific facts to support and flesh out these factual assertions, defendants moved under § 1867(f) for an order requiring disclosure for their inspection and copying of certain records and papers used in the jury selection process. Among the items specifically requested were forms allegedly filed by district court personnel that would reflect the percentages of racial minorities on the jury lists in recent use.
The district court denied the motion for production of the records and papers on the basis that the defendants’ motion papers failed to establish a prima facie case of violation of the fair-cross-section requirement of the Act and the Constitution, and that this was a prerequisite to their right to disclosure. The court then denied on the merits the motion to dismiss the indictments, holding that the defendants’ motion papers conclusively demonstrated that they could not prove a prima facie case of violation under Duren v. Missouri, 439 U.S. 357, 364, 99 S.Ct. 664, 668, 58 L.Ed.2d 579 (1979).
On this appeal, the government has conceded, as it must, see Test v. United States, 420 U.S. 28, 95 S.Ct. 749, 42 L.Ed.2d 786 (1975), that the district court erred in denying the disclosure motion. The defendants’ factual assertions in support of their motion were sufficient to confer the “unqualified right” to inspect and copy relevant records that is provided by § 1867(f). They need not go further in proof of their claim to establish this right. See Test, 420 U.S. at 30, 95 S.Ct. at 750.
But the government then contended that this error was harmless because, as the district court further concluded, defendants’ motion papers conclusively demonstrated that defendants could not prove a violation of the Act or the Constitution. Specifically, says the government, as the district court correctly held, even if significant underrepresentation of racial and age groups on the jury lists were established, this would not prove the “systematic exclusion” of “distinctive groups” which the Act and the Constitution forbid. “Systematic exclusion,” urges the government, means intentionally discriminatory exclusion of distinctive groups. Here, defendants have only suggested that substantial underrep-resentation results from exclusive reliance upon voter registration lists; they have not suggested intentional discrimination either in denying access to the voter registration process, or in compiling the jury lists. Thus, on the defendants’ own allegations, any underrepresentation that could be shown would only be chargeable to the inaction of prospective jurors in failing to register, not to any governmental discrimination. Therefore, accepting all of the defendants’ assertions, whether of fact or of probabilities respecting underrepresentation, no violation of the fair-cross-section requirement of the Act or Constitution would be proven.
In sum, the government has argued on appeal that even substantial underrepre-sentation of distinctive groups that is traceable solely to exclusive reliance on voter registration lists from which members of those groups have freely excluded themselves cannot constitute a violation of the fair-cross-section requirement. Or, put differently, that exclusive reliance upon voter registration lists is per se compliance with the fair-cross-section requirement where no intentional discrimination in composing either voter or jury lists is shown.
As I understand the court’s holding today, it essentially adopts this legal position as the primary ground for its affirmance of the district court’s dismissal of the jury challenge, see majority op. 1444-51. It then holds alternatively that, in any event, the record on appeal, fleshed out by taking judicial notice of various official data, conclusively demonstrates that no group identified by defendants is both “distinctive” and “substantially underrepresented” on the relevant grand or petit jury venires. Majority op. 1451-55.
I disagree with both of these alternative grounds for affirmance.
On the much more critical legal ground, I believe that under controlling Supreme Court authority any “substantial underrep-*1462reservation” of “distinctive groups” on grand or petit jury venires incident to the government’s reliance upon voter registration lists as its exclusive source for composing juries violate the fair-cross-section requirement of the Act and the sixth amendment. I think this is so even though the government has not intentionally discriminated against members of the group in either the voter registration or jury selection processes, and though underrepre-sentation might be traceable ultimately only to the voluntary failure of members of the distinctive group to register as voters in sufficient numbers.
On the alternative ground, I think that the facts are not sufficiently developed on the meager record we review, even when supplemented by judicial notice of the data relied upon by the court, to resolve the critical factual issues, never addressed by the district court, as to whether there may be substantial underrepresentation of at least one “distinctive group” among those identified in the defendant’s motion.
I take the legal and factual bases in order.
A
The court’s rationale for its legal position —a rationale apparently adopted by a number of other courts18 — focuses on the fact that, absent discriminatory causes, under-representation on jury lists that traces to “underrepresentation” on voter registration lists is chargeable only to a free choice of those who have failed to register. Such underrepresented groups cannot therefore be considered “cognizable” under the policies that underlie the Act and, behind the Act, the fair cross-section requirement of the sixth amendment. See Camp v. United States, 413 F.2d 419, 421 (5th Cir.1969) (Jehovah’s Witnesses “who do not choose to register” not a “cognizable” group).
I simply disagree with that rationale,19 and believe that intervening Supreme Court decisions flatly undercut it.
Starting with Camp, the stated rationale of these decisions and now of this court is that it is not unfair — hence does not violate either statutory or constitutional “fair cross-section” requirements — for govern*1463ment to exclude persons from jury service for reasons traceable solely to those persons’ own uninhibited choices. A passage from the Act’s legislative history may be the source of the rationale, see United States v. Armsbury, 408 F.Supp. 1130, 1140 (D.Ore.1976); certainly it expresses the same notion.
In a sense the use of voter lists as the basic source of juror names discriminates against those who have the requisite qualifications for jury service but who did not register or vote. This is not unfair, however, because anyone with minimal qualifications — qualifications that are relevant to jury service — can cause his name to be placed on the list simply by registering or voting. No economic or social characteristics prevent one who wants to be considered for jury service from having his name placed in the pool from which jurors are selected.
H.R.Rep. No. 1076, 90th Cong., 2d Sess. (1968), reprinted in 1968 U.S.Code Cong. & Ad.News 1794-95.
The problem with this view as a basis for decision — whether as an assumed direct expression of legislative intent or as an independently derived rationale — is that it relates to only one of the policies behind the Act, and not at all to the central sixth amendment right that the Act seeks to implement.
The Act expressly identifies two policies that it seeks to implement: (1) that all litigants “shall have the right to grand and petit juries selected at random from a fair cross section of the community,” and (2) that “all citizens shall have the opportunity to be considered for [jury] service.” 28 U.S.C. § 1861.
The quoted passage of legislative history and the rationale of the Camp line of decisions and now of this court have obvious relevance to the second policy — that of securing to “all citizens” a fair opportunity to serve on juries. Therefore, if viewed only from the perspective of the intended beneficiaries of that policy — prospective jurors —it makes sense to hold that when they have excluded themselves, there has been no violation of that policy. But this passage of legislative history and the parallel rationale of the Camp line of decisions have no relevance to the first policy — that of securing to criminal defendants the right to be charged and tried by jurors randomly chosen from a fair cross section of the community. Consequently, the mere fact that the second policy may not be violated when underrepresentation results from self-exclusion does not mean that the first policy may not be.
That this is so is indeed reflected in other portions of the Act’s legislative history that focus on the policy of securing the rights of criminal defendants respecting juror selection processes rather than the correlative but different rights of prospective jurors to serve. The Act’s policy respecting defendants’ rights implements and directly mirrors the fair cross-section requirement rooted in the sixth amendment. See Taylor v, Louisiana, 419 U.S. 522, 528-30, 95 S.Ct. 692, 696-98, 42 L.Ed.2d 690 (1975); United States v. Herbert, 698 F.2d 981, 984 (9th Cir.1983); United States v. Test, 550 F.2d 577, 584 (10th Cir.1976); United States v. Whiting, 538 F.2d 220, 222 (8th Cir.1976). Reflecting this, the legislative history explains that voting lists “need not perfectly mirror the percentage structure of the community,” but nevertheless admonishes that:
any substantial percentage deviations must be corrected by the use of supplemental sources. [H.R.Rep. No. 1076, at 1794.]
* # * * # #
Jury performance will be enhanced as well by closer approximation of the cross sectional goal under the bill. It must be remembered that the jury is designed not only to understand the case, but also to reflect the community’s sense of justice in deciding it. As long as there are significant departures from the cross sectional goal, biased juries are the result-biased in the sense that they reflect a slanted view of the community they are supposed to represent. [Id. at 1797.]
* * * * * #
*1464[I]n accord with section 1863(b)(2), voter lists will be supplemented by other sources of names if they do not reflect a community cross section. [Id. at 1798.]
* sjc * * # *
[M]oreover, the bill recognizes that in some areas voter lists of all kinds may be insufficient to implement the policies of the act, by reason of local voting practices. Where that is true, the plan must prescribe other sources to supplement the voter lists. [Id. at 1799 (emphasis added).]
These excerpts from the Act’s legislative history conclusively demonstrate a legislative intent that underrepresentation resulting from exclusive reliance on voter registration lists might violate the Act even though traceable ultimately to self-exclusion by free choice of members of underrepresented groups.
More critically, Duren v. Missouri, 439 U.S. 357, 99 S.Ct. 664, 58 L.Ed.2d 579 (1979), elaborating on Taylor v. Louisiana, 419 U.S. 522, 95 S.Ct. 692, 42 L.Ed.2d 690 (1975), buttresses if it does not compel this reading of the statute by its analysis of the underlying sixth amendment right.20 Addressing a sixth (and fourteenth) amendment fair-cross-section challenge to a criminal defendant’s state court conviction, the Court considered the effect of underrepre-sentation of women on jury lists resulting from their self-exclusion under a state law exemption privilege. Assuming that self-exclusion was the effective cause of under-representation, and concluding that no significant justification for the exemption-privilege had been shown, the Court held that “exclusion” by this process was “systematic.” Id. at 366, 99 S.Ct. at 669. Because the degree of underrepresentation was also found constitutionally unacceptable (less than 15% represented out of 54% eligible), id., the Court found the sixth amendment fair-cross-section requirement violated.
Duren thus undermines the critical assumption made by this court that the fair-cross-section requirement only protects against intentional discrimination in the jury selection process; that exclusion of distinctive groups from jury lists is only “systematic” if it results from intentional discrimination. Rather, as Duren expressly recognizes, it protects more broadly against any process of jury selection which produces jury master lists on which “distinctive groups” are substantially underrepresented for any reasons “inherent” in the process used by government to constitute the lists, i.e., other than the vagaries of chance in a random selection process. Id. at 366, 99 S.Ct. at 669.21 Specifically, on the direct authority of Duren and Taylor, it is now clear that “systematic exclusion” may be found where substantial un-derrepresentation results from self-exclusion by members of the distinctive groups when that self-exclusion is permitted by — is thus “inherent in” — a challenged process.
Therefore, just as the permitted self-exclusion of women leading to their underrep-*1465resentation on jury lists was found to violate the constitutional fair-cross-section requirement in Duren and Taylor, so the alleged underrepresentation of distinctive groups by their self-exclusion from exclusive-source voter registration lists might be found here to violate the fair-cross-section requirement of the Constitution and the Act.22
B
In its fail-back position, the court holds that even if substantial underrepresentation traceable to exclusive reliance on voter registration lists could constitute a statutory and constitutional violation, the record here, fleshed out by judicial notice, conclusively shows that there is no substantial underrepresentation of any of the distinctive groups tentatively identified by defendants. It may well be that this is the ultimate fact of the matter. Indeed my own instinct is that it quite probably is. But I do not think that under the circumstances we should attempt to make that determination on this appeal. The district court has never, of course, addressed the specific question of statistically significant underrepresentation. On the meager factual record before it, it could not have. This court has now sought to flesh out the record by taking judicial notice of certain census data concerning national, regional and Maryland voter registration percentages involving blacks and whites (and in a most amorphous way, different age groupings). The court confidently asserts that the data it has selected is the appropriate data and that it conclusively demonstrates that none of the identified groups are substantially underrepresented.
Again, my instinct is that the court selected data, at least that related to black and white voter percentages in the Maryland district as of November 1982, is probably the most appropriate available data concerning voter registration figures for those groups and that it is accurate. But this data has never been subjected to adversarial party scrutiny. The course of proceedings in both the district court and this court have provided no opportunity for that. And, even more critically, all the data stops at the level of voter registration percentages. While it may be assumed, as the court obviously does, that there has been no contrivance or slip-up in converting voter registration lists directly to master jury lists, this can only be assumed by this court. The challenge here is not — as the court continually seems to believe — to un-derrepresentation on voter registration lists or to the use vel non of those lists as the exclusive source of master jury lists. The challenge — at this stage avowedly tentative — is to underrepresentation on the master jury lists from which jury venires and panels are required to be drawn randomly. This, not corruption of the voting franchise, is what the Act is designed to guard against. 28 U.S.C. § 1861 (policy declaration). And it is data showing the ultimate composition of master jury lists and venires that the Act allows defendants to discover in order to perfect their tentative challenges. 28 U.S.C. § 1867(f). The critical datum, therefore, is that showing *1466the representation of any groups determined to be “distinctive” on the master jury lists at the critical time. That datum is simply not in the record even as expanded, and although it is supposed perfectly to match the voter registration lists, that can only be assumed on the present record.
Even though the risk of factual error in making the underrepresentation determination on this judicially expanded record may be thought minimal, it is clearly present. In my view, with the case at this stage, considerations of delay and judicial economy simply do not justify running that risk. I would therefore vacate the judgments and remand to the district court for reconsideration of the jury selection challenge after defendants have been allowed to inspect, reproduce, and copy the jury selection records whose disclosure they have sought and erroneously been denied. See Test v. United States, 420 U.S. 28, 30, 95 S.Ct. 749, 750, 42 L.Ed.2d 786 (1975); United States v. Lawson, 670 F.2d 923, 930 (10th Cir.1982); United States v. Marcano-Garcia, 622 F.2d 12, 18 (1st Cir.1980). This would obviously not require wholesale reversal of the convictions. Jurisdiction should be retained in this court for the limited purpose of reviewing the district court’s ruling on the properly considered challenge. If the district court denied the challenge, it could be authorized simply to reinstate the judgments of conviction (except that of Cecil). See Marcano-Garcia, 622 F.2d at 18.23
Ill
At this point it seems appropriate to address the parade of practical evils that the court suggests must attend all underrepre-sentation challenges that are not confined narrowly to intentional discrimination. See majority op. 1454-1455. These are apparently advanced both to support the court’s narrow reading of the permissible basis of challenge intended by the Act, and possibly to justify the court’s venture into appellate fact-finding in order to spare the district court that function in this case.
With all respect, the court’s assumptions and fears are considerably overwrought.
In the first place, the possibility of such challenges in any criminal case has been expressly created by Congress, as an integral part of the Act’s approved plan regime. It is obvious that the predominant purpose is to provide a means for questioning whether approved plans relying exclusively upon voter registration lists are actually providing the fair cross section of jurors that the Act is designed to insure. That they may not over time continually do so, for whatever reason, is obvious, and Congress clearly assumed that they might not always do so. The procedure for mounting such challenges is not therefore, as the court seems to suggest from time to time, a contrivance of the defendants in this case, but is a safeguard deliberately designed by Congress to justify primary reliance on voter registration lists. We obviously may not negate a procedure provided by Congress because of concerns about its possible abuses.
Being available, the procedure is certainly subject to abuse by irresponsible defense counsel and the court here is obviously much concerned with that possibility. But *1467there is ready practical protection against that risk which, if effectively used, will substantially minimize the dread consequences that the court envisions.
This protection lies in effective advocacy by government counsel and effective management of the challenges by district courts. If these occur, not only may non-meritorious specific challenges be summarily (but properly) disposed of, but the scope of future challenges may be systematically narrowed by stare decisis.
In meeting an underrepresentation challenge to the jury array, the government may always of course deny at the threshold that particular groups identified as “distinctive” by the challenge are indeed so. This will present an issue that can ordinarily be determined as a matter of law under well-developed legal principles. For example, the First Circuit has now, by that process, pretty effectively narrowed to racial minorities and women the groups that will be considered cognizable in that circuit, specifically ruling out groups based on age, educational level, and economic class.24 As a matter of fact, the District Court for the District of Maryland has earlier held that “age alone does not define a distinct group,” United States v. Blair, 493 F.Supp. 398, 406 (D.Md.1980), aff'd on other grounds, 665 F.2d 500 (4th Cir.1981), and that precedent could of course have been directly applied here to reject the 18-34 age group challenge.
Beyond this threshold means of summarily rejecting claims for lack of group cog-nizability, there is an almost equally ready means for summarily but properly rejecting nonmeritorious claims of substantial underrepresentation of admittedly “cognizable” groups. As indicated earlier, the critical datum on underrepresentation is the proportional representation of cognizable groups on the relevant master jury lists and venires. There is no apparent reason why this datum cannot be presented quickly and decisively by government counsel to demonstrate that exclusive reliance on voter registration lists is not producing substantial underrepresentation of particular groups on the relevant jury lists and venires. To fail to do so on the basis that the challengers have the burden to prove underrepresentation would seem the height of folly. Indeed it would seem no great administrative inconvenience for the persons charged with administration of approved jury plans to maintain on a continuously updated basis the data as to those groups clearly established as “cognizable” under controlling precedent. By the development of precedent, these groups may well be narrowed fairly quickly (possibly, as in the First Circuit, essentially to women and the dominant ethnic minority). Under such an administrative regime, which might well be specifically directed by district courts, this evidence would be readily available to government counsel for quick and effective use to forestall nonméritorious underrepresentation claims at the threshold.
As the majority points out, experience is demonstrating the wisdom of Congress’ authorization of primary reliance on voter registration lists as the source of master jury lists and ultimately of venires. This should work and it demonstrably is working. But that the national experience has demonstrated the wisdom of the legislative judgment does not of course confer a general immunity to specific challenges of specific district plans. The answer in individual districts lies rather in wise utilization by government counsel and the district courts of the ready means for demonstrating that the national experience holds true for the district.25
A.
In sum, I disagree most critically with the majority’s primary holding that the Act and the underlying sixth amendment fair cross section requirement protect only against intentional discrimination. I be*1468lieve instead, that the Act, implementing the sixth amendment, goes beyond the prohibition of intentional discrimination and requires that where substantial underrep-resentation exists by virtue of exclusive reliance on voter registration lists, affirmative measures must be taken to insure realization of the fair cross section requirements. See United States v. Jenkins, 496 F.2d 57 (2d Cir.1974).
Beyond this, I also disagree with the court’s effort to decide on this appeal that in fact there is no substantial underrepre-sentation of any group suggested by defendants. That determination is one properly made in the first instance by the district court, which has not yet addressed it, and which could now determine it without undue burden and without significant disruption of the progress of this case to its conclusion.
Judge WINTER and Judge MURNA-GHAN have asked to be shown as joining in this separate opinion.
. It is not clear from the record just when Cecil's counsel advised government counsel about Creswell’s aborted insanity plea in the New York proceedings, but it was obviously not a bombshell dropped in mid-trial as the majority implies. Majority op. 1434. In Cecil's brief on this appeal, it is asserted that notice was given an Assistant United States Attorney by letter back in 1981 (even before indictment), Appellant’s Br. 10, and this is nowhere questioned. The record itself shows that Cecil’s counsel had provided the government with copies of the New York insanity plea proceedings approximately a week before Creswell began to testify, J.A. 251. In any event, and for whatever relevance it has, it is not accurate to suggest, as the majority opinion does, at 1434, that the evidentiary issue was first "raised" by Cecil during trial, and that the government only made its motion in limine in response to this sudden development. Actually, of course, Cecil did not “raise" the admissibility issue with the court; the government did, on the basis of information sometime earlier supplied it by Cecil’s counsel.
. The majority points out that after all the evidence, including Creswell’s testimony, was in, Cecil sought to be allowed to call the psychiatrist as a witness, or, failing that, to put his "certification” in evidence; and the majority for some reason then questions whether the psychiatrist was available even then. In total context, it is obvious that at that late point, having been denied the critical right to cross-examine Cres-well, Cecil’s counsel was simply protecting the record. The district court had already ruled that no evidence, whether by cross- or direct examination, on the matter of Creswell’s insanity plea notice would be admitted.
. A succession of scope-limiting proffers appear at each of the referenced pages of the Joint Appendix. The final comprehensive one, which may be taken as the subject of the court’s ultimate exclusionary ruling, was in these words:
I am going to ask him whether or not he participated in his defense with his lawyer. Whether he was concerned with the case against him. Whether he saw the correspondence that was sent in his behalf. The declaration is filed in his behalf. Whether he consulted and agreed with the activities of his lawyer on his behalf. Whether he entered or attempted to enter a plea of not guilty by reason of insanity. Whether he felt that he had a pathological disorder at that time that he discussed the matters with the doctor in 1981, and prior — that is, in 1980, at the time of the alleged offense. J.A. 267.
. This appears most plainly from a closing colloquy between the court and Cecil’s counsel, Horn, following the district court’s final ruling that "I will exclude the matter of claimed psychiatric background of this defendant [sic] in the course of his cross-examination." J.A. 290-91. At that point, the following exchange occurred:
MR. HORN: Judge, again, just for clarification, does that preclude me from asking whether he joined in the attempt to enter a not guilty by reason of insanity plea?
THE COURT: Yes.
. The rationale goes back at least to Camp v. United States, 413 F.2d 419 (5th Cir.1969), and has at some time apparently been followed by the courts in United States v. Evans, 542 F.2d 805, 812 (10th Cir.1976) (asserting that “the circuits are in complete agreement" on the proposition); United States v. Freeman, 514 F.2d 171, 173 (8th Cir.1975) (cited by government) (Indian underrepresentation by "their own indifference” provides no basis for challenge); United States v. Lewis, 472 F.2d 252, 255-56 (3d Cir.1973) (cited by government) (no violation where underrepresentation of blacks is "result of their own inaction”); United States v. Guzman, 468 F.2d 1245, 1248 (2d Cir.1972) (no supplementation required unless “obstacles [to register] are placed in the paths” of underrepresented group members); United States v. Ross, 468 F.2d 1213, 1216 (9th Cir.1972) (no violation where no showing that "right of young persons to register ... has been inhibited in any way”).
. As do other federal courts. See, e.g., the following decisions in which, without according presumptive validity to plans relying exclusively on voter registration lists, courts have analyzed claims of underrepresentation solely traceable to that factor: United States v. Pepe, 747 F.2d 632, 648-49 (11th Cir.1984); United States v. Jenkins, 496 F.2d 57, 64-66 (2d Cir.1974); Waller v. Butkovich, 593 F.Supp. 942, 948-58 (M.D.N.C.1984). See also United States v. Armsbury, 408 F.Supp. 1130, 1139-40 (D.Ore.1976) (expressly rejecting Camp rationale); United States v. McDaniels, 370 F.Supp. 298, 301-12 (E.D.La.1973) (same; semble), aff'd sub nom. United States v. Goff, 509 F.2d 825 (5th Cir.1975).
Interestingly, the recent First Circuit cases heavily relied upon by the majority op. 1445-1446, as supporting its position, Barber v. Ponte, 772 F.2d 982 (1st Cir.1985) (en banc); Anaya v. Hansen, 781 F.2d 1 (1st Cir.1986); and United States v. Lynch, 792 F.2d 269 (1st Cir.1986), do not do so, but to the contrary support the position taken in this dissent. Those decisions merely held that a number of group characteristics — age, type of work, educational level — do not make the relevant groups "cognizable" or "distinct” for fair cross-section underrepresentation purposes, unless made so by specific "invidious" governmental discrimination. Anaya, 781 F.2d at 7. But the decisions at the same time expressly recognize that if a group is intrinsically "distinct” or “cognizable" (such as women and racial minorities), a simple showing of significant statistical underrepresentation without regard to cause, constitutes prima facie proof of a violation of the fair cross-section requirement of the Act or the sixth amendment, Barber, 772 F.2d at 998; Anaya, 781 F.2d at 7; see also United States v. Marcano-Garcia, 622 F.2d 12 (1st Cir.1980) (viable underrepresentation claim recognized without proof of intentional discrimination).
. Duren was decided after the last of the Camp line of decisions cited in note 18, supra. For that reason, it may be questioned whether those decisions are still considered authoritative on the issue by the rendering courts. Against this possibility, it must be conceded that some of the courts have, since Duren, continued to rely upon pre-Duren decisions applying the Camp rationale. See, e.g., Brown v. Lockhart, 781 F.2d 654, 656 (8th Cir.1986) (citing United States v. Freeman, 514 F.2d 171); United States v. Afflerbach, 754 F.2d 866, 869-70 (10th Cir.1985) (citing, inter alia, United States v. Lewis, 472 F.2d 252).
As can be seen from these cases and those collected in notes 18 and 19, the lower federal courts are probably simply in conflict on the issue.
. Duren, incidentally, identifies the different constitutional sources of the dual policies of the Act, hence the different constitutional tests for violation of the two, and in that process also probably identifies the source of the flawed Camp -line rationale. Pointing out that the equal protection right of citizens to serve on juries is protected only against intentional denials of the right, the Court distinguished that right (as it might be directly or derivatively asserted) from the sixth amendment fair-cross-section right of defendants. The latter right, the Court noted, is violated by "systematic disproportion itself.” Duren, 439 U.S. at 368 n. 26, 99 S.Ct. at 670 n. 26.
The Camp rationale is thus shown to be apposite only to the equal protection right, not to the sixth amendment right, both of which are sought to be implemented by the Act.
. The majority says that Duren does not control decision here because there the self-exclusion occurred at the jury selection stage rather than, as here, at the voter registration stage. Majority op. 1449-1451. With all respect, this is a distinction without a difference. Where the governmental process used is one in which the voter registration list is exactly the same as the master jury list, self-exclusion from the one is directly self-exclusion from the other. Here, as surely as in Duren, the phenomenon by which the potential master list of potential jurors could be reduced by a process of permitted self-exclusion is "inherent” in the system by which a particular jury venire is ultimately composed. That the privilege of direct exemption from jury service was expressly authorized by statute in Duren while the privilege not to register as a voter simply results from the absence of governmental compulsion to exercise the voting franchise here — is equally a distinction without a difference.
On this point the majority cites, see majority op. 1450, 1451, the present Chief Justice’s comment in denying a stay in California v. Harris, 468 U.S. 1303, 105 S.Ct. 1, 82 L.Ed.2d 807 (1984), as making "clear” the correctness of its view. With all deference, I believe a more suitably cautious reference would emphasize the Chief Justice’s remark that the issues on which we on this court now disagree “are by no means open and shut questions under Duren" Id. at 1304, 105 S.Ct. at 1.
. The remand I propose would of course express no opinion on the ultimate merits of the jury selection challenge. It would simply direct consideration of the challenge under the proof scheme outlined in Duren. Under that scheme a challenge establishes a prima facie violation by showing that (1) there is a distinctive group (or groups) in the community (2) which is substantially underrepresented on relevant jury ve-nires in relation to its numbers in the community (3) due to systematic exclusion of group members in the jury selection process. Duren, 439 U.S. at 364, 99 S.Ct. at 668.
The holding I propose in this dissent would only have decided prong (3) in this scheme: that any underrepresentation of distinctive groups traceable to exclusive reliance on voter registration lists would be "systematic exclusion.”
Even if a prima facie violation were shown, the process might yet be found not to violate the fair cross section right if the government proved that "attainment of a fair cross section [is] incompatible with a significant state interest.” Id. at 368, 99 S.Ct. at 670.
This is obviously business for trial courts rather than appellate courts attempting to act in the first instance on records largely constructed by unchallengeable judicial notice.
. Barber v. Ponte, 772 F.2d 982 (1st Cir.1985) (en banc); Anaya v. Haman, 781 F.2d 1 (1st Cir.1986); United States v. Lynch, 792 F.2d 269 (1st Cir.1986).
. Of course, there is another safeguard against flagrant abuses of the challenge procedures: the inherent power of the district courts to discipline attorneys.