Benjamin J. Butera was indicted for attempted income tax evasion by the federal grand jury for the Southern Division of Maine in March 1968. On April 8, he filed with the district court a motion to dismiss the indictment on the ground that it had been returned by a grand jury drawn from an improperly constituted jury pool. A hearing was had and evidence heard, and defendant’s motion was denied. Sub. nom. United States v. Bryant, 291 F.Supp. 542 (D. Me. 1968).1 Defendant was subsequently found guilty of the offense charged. He brings this appeal to contest the denial of his motion to dismiss the indictment.

*567Defendant contends that the grand jury which returned his indictment was improperly constituted because certain segments of the population of southern Maine were inadequately represented on the jury pool from which his grand jury was drawn. More specifically, he claims underrepresentation of (1) the young and the very old, (2) women, (3) the less educated, and (4) people of certain Maine counties. He concedes that proportional representation is not required; he also concedes that the disparities did not arise from any purposeful or deliberate discrimination in the sense that the jury pool selection system was administered with a lack of good faith. He claims simply that the system used in Maine resulted in unconstitutional discrimination against members of what he claims are legally cognizable groups.2

The Constitutional Mandate

This contention necessitates a determination of what the Constitution requires concerning the selection of juries. The Supreme Court has consistently required that jury selection systems draw their jurors from a fair cross section of the community.3 It has been suggested that such non-discriminatory jury selection is an essential aspect of our democratic form of government.4 However, the Court has long recognized that fair and reasonable qualifications for jury service eligibility can be imposed even though they detract from a cross section in the actual jury pools.5 Moreover, the Court has recognized that it is neither possible nor necessary — in order to assure an impartial jury — that there be a fair cross section of the community on each individual grand and petit jury.6 In particular, Hoyt v. Flori*568da, 368 U.S. 57, 82 S.Ct. 159, 7 L.Ed. 2d 118 (1961), demonstrates the Court’s conviction that an impartial jury can be assured even though members of one’s class are not present either on the jury or in the jury pool. In that case, despite the fact that the defendant — who had killed her husband with a baseball bat — was tried by an all-male jury drawn from a virtually all-male jury pool, the Court upheld the conviction because there had been no impermissible discrimination against women in the operation of the jury system.

We conclude from these cases that the Supreme Court has focused on whether the alleged underrepresentation in the jury pool is the result of discrimination in the juror selection system. Coneededly, most of the jury exclusion cases have involved alleged state discrimination and thus the application of the Fourteenth Amendment’s assurance of equal protection. Here we deal with a federal jury and can exercise our supervisory power in addition to the Constitutional requirements.7 However, even in exercising this supervisory power, the Supreme Court has focused on discriminatory selection practices. Ballard v. United States, 329 U.S. 187, 195, 67 S.Ct. 261 (1947); Thiel v. Southern Pacific Co., 328 U.S. at 220, 225, 66 S. Ct. 984. We therefore understand the Constitutional mandate to be for a jury selection system free of discrimination.

In carrying out this mandate, the Supreme Court has normally imposed on the defendant the initial burden of demonstrating, prima facie, the existence of purposeful discrimination.8 However, the exact meaning of “purposeful discrimination” has been elusive at best. Such discrimination has been found when the jury commissioners limited their selections to acquaintances or certain lists of persons which necessarily resulted in discrimination against a class of persons not falling within either category.9 It has been found when Negroes were underrepresented and the list or method used indicated each prospective juror’s race, thereby providing the opportunity for discrimination.10 Sometimes it has been found when complete exclusion or significant disparities existed which could not be adequately explained or justified by the responsible officials.11 At other times it has been *569found when jury commissioners, albeit completely well-intended, deliberately excluded certain groups.12

Thus, while “purposeful discrimination” may connote an element of bad faith in ordinary usage, the term has not been so limited by the Supreme Court; rather, the breadth with which the term has been used by the Court indicates that purposeful discrimination exists whenever significant unexplained disparities exist. In other words, it is not the significant disparities themselves which are unconstitutional, Akins v. Texas, 325 U.S. at 403-404, 65 S.Ct. 1276; Hoyt v. Florida, 368 U.S. at 69, 82 S.Ct. 159; they only raise the inference of discrimination. E. g., Billingsley v. Clayton, 359 F.2d 13, 17 (5th Cir. 1966), (en banc), cert. denied, 385 U.S. 841, 87 S.Ct. 92, 17 L.Ed.2d 74 (1966); Witcher v. Peyton, 382 F.2d 707, 709-710 (4th Cir. 1968); Salary v. Wilson, 415 F.2d 467, 470-471 (5th Cir. 1969). Once that inference has been raised, it is the government’s failure or inability to demonstrate that the disparities are not the product of discrimination which confirms the inference and invalidates the jury pool. E. g., Witcher v. Peyton, 405 F.2d 725 (4th Cir. 1969). Finally, oúr reading of the cases indicates that underrep-resentation of the more sociologically distinct groups in our country necessitates a more compelling demonstration by the government to overcome the inference of discrimination. Compare Hoyt v. Florida, 368 U.S. 57, 82 S.Ct. 159 (1961) with Whitus v. Georgia, 385 U.S. 545, 87 S.Ct. 643, 17 L.Ed.2d 599 (1967).

Defendant’s Prima Facie Case

Defendant has the burden of showing, prima facie, discriminatory selection practices. Whitus v. Georgia, 385 U.S. at 550-551, 87 S.Ct. 646, 647. Equally important, he has to show such discrimination against “distinct” groups in the general population. Hernandez v. Texas, 347 U.S. at 478-480, 74 S.Ct. 667. We think he has disclosed sufficient disparities, in three cognizable groups, to raise a reasonable inference of discrimination which the government should come forth to dispel.

Defendant’s first claim relates to age; he insists that the young and very old were not adequately represented on the jury pool. Obviously there are some significant disparities in the age groups as defendant has classified them.13 *570Equally obvious, we think, is the problem of deciding what age groups are cognizable for purposes of defendant’s prima facie ease. See King v. United States, 346 F.2d 123, 124 (1st Cir. 1965). We do not believe that that question can be given a definitive answer which will be satisfactory for all cases; rather, we think that age groupings must be viewed from several perspectives in order to give a fair view of the significance of any disparity.

We take defendant’s breakdown of age groups as one acceptable perspective. More importantly, we note that there were only 148 persons under age 40 when defendant’s “true cross section”, see n. 13, indicated 335, and that there were only 20 persons under age 30 when the “true cross section” indicated 147. Below age 35, defendant’s three age groupings show minimal representation of the jury pool; above age 35, the representation appears satisfactory and raises no inference of discrimination.14 These observations, when taken together, indicate to us a sufficient disparity to infer — in the absence of some explanation and justification — some discrimination against young adults. Moreover, the government has not directed our attention to any jury qualification or exemption which would on its face adequately explain this underrepresentation of young adults.15

Finally, we are satisfied that young adults constitute a cognizable— though admittedly ill-defined — group for purposes of defendant’s prima facie case. We cannot allow the requirement of a “distinct” group to be applied so stringently with regard to age grouping that possible discrimination against a large class of persons — in our case, those between 21 and 34 — will be insulated from attack. Nor can we close our eyes to the contemporary national preoccupation with a “generation gap,” which creates the impression that the attitudes of young adults are in some sense distinct from those of older adults. That apparent distinctness is sufficient for us to say that neither class could be excluded from jury pools without some justification. Accordingly, we find the “significant disparity” with regard To age which raises the inference of discrimination and shifts the burden of explanation to the government.

Defendant’s second claim relates to an alleged discrimination ac*571cording to sex; his figures indicate that the actual jury pool was only 36 per cent female when the “true cross section” was about 52 per cent female. While this class poses no problem of definite identification, the disparity is not so great as it was with regard to age. However, a 30 per cent underrepresentation of so large and important a group as women cannot lightly be dismissed; it at least raises an inference of discrimination. Unlike the situation in Hoyt v. Florida, 368 U.S. 57, 82 S.Ct. 159 (1961), there is no statutory provision to explain the disparity. That males and females can have sufficiently different points of view to be considered legally cognizable groups for purposes of defendant’s prima facie ease should not require the authority which we nevertheless cite. Ballard v. United States, 329 U.S. at 193-194, 67 S.Ct. 261 (Douglas, J.).

Defendant’s third claim relates to educational attainment; he insists that his statistics indicate an impermissible un-derrepresentation of the less educated.16 Expressed in another manner, they indicate that 83.6 per cent of the actual jury pool had at least a high school diploma, while only 43.2 per cent of the general population of Maine over age 25 did; 18.1 per cent of the actual jury pool had at least a college degree, compared with 5.5 per cent of the general population.

It seems clear that these disparities give rise to an inference of discrimination against the less educated, despite the difficulty, similar to that encountered with regard to age groupings, of defining any precise group to be termed “less educated”. Again, to require a more precise definition would introduce unnecessary and , unrealistic inflexibility and might effectively preclude anyone from ever showing a “distinct” class in terms of educational attainment. Moreover we think the statutory requirement of literacy on its face explains only a part of the variance; there is no statutory provision for “blue ribbon” juries which might explain the remainder. Fay v. New York, 332 U.S. 261, 67 S.Ct. 1613 (1947). Finally, we are convinced that the less educated are a sufficiently large group» with sufficiently distinct views and attitudes that its diluted presence on the actual jury pool requires some explanation by the government. See Thiel v. Southern Pacific Co., 328 U.S. at 220, 223-224, 66 S.Ct. 984.

Defendant’s fourth contention gives us more difficulty. His figures indicate some disparities between the actual jury pool and the general population of southern Maine, in terms of county residence.17 First of all, these are generally not as serious discrepancies as those which have appeared above. Secondly, we note that the relevant statute provides that jurors can be selected with regard to their distance from the courthouse, 28 U.S.C. § 1865(a) (1964), the validity of which has not been at*572tacked by defendant. See Katz v. United States, 321 F.2d 7, 8-9 (1st Cir. 1963), cert. denied, 375 U.S. 903, 84 S.Ct. 193, 11 L.Ed.2d 144 (1963). At least part of these rather minor disparities may be attributable to this statute. More importantly, however, we are not aware that residents of counties can be said to hold views and attitudes which are in any way “distinct” from those of their neighbors in nearby counties, nor has defendant given us any evidence of such distinctness. While common experience tells us that people’s attitudes differ to some degree along lines of age, sex and extent of education, we are not aware that they differ along county lines. We have been willing above to give a broad meaning to the requisite “distinctness” of classes but in each instance we could point to some indication that the groups isolated by defendant — at least in a general sense — possessed the essential element of distinctness.18 That term would have no meaning at all were we to say— in the absence of any supporting evidence — that residents of some counties have views and attitudes genuinely distinct from those of nearby counties. Cf. Krause v. Chartier, 406 F.2d 898, 901 (1st Cir. 1968) (drawing of jurors from last names T-Z not unconstitutional).

The defendant having raised an inference of discrimination against young adults, women, and the less educated, the burden shifts to the government to dispel that inference. It should be remembered at the outset that, while a true cross section is the ultimate ideal, it is by no means the Constitutional mandate. What is required is a jury selection system free of discrimination against properly cognizable groups.

The Government’s Explanation

The government offered the testimony of one of the Jury Commissioners — the other having died prior to the hearing— as well as various documents evidencing the manner in which the “key man” system was administered in Maine during the 1960’s. A full explanation of the system appears in United States v. Bryant, D.C., 291 F.Supp. at 544-546. Briefly, the two Jury Commissioners, working in conjunction with the district court, first determined that about 900 persons were needed for the jury pool in the Southern Division. They then determined how many of those 900 should be drawn from each community by ascertaining the total number of registered voters in southern Maine, dividing that number by 900, finding that about one of every 275 registered voters would be needed, and then allocating a rough quota to each local community based on that proportion. Thereafter, they obtained through the Maine Register the names of several hundred “key men”, most of whom were local town officials.19 These key men were asked by letter to send the names of four or five potential jurors to the Jury Commissioners, who then sent questionnaires to the persons recommended. After receiving the completed questionnaires, the Jury Commissioners struck nearly half for reasons of statutory ineligibility or apparent hardship and placed the remaining names in *573the jury pool for selection at random of' a grand jury.

Defendant concedes that the key man system is not per se invalid.20 From the evidence adduced before the district court, it appears that the Jury Commissioners acted at all times in good faith— defendant concedes that — and in conscientious accordance with the applicable statutes and the suggestions gleaned from the 1960 Judicial Conference Report, 26 F.R.D. 409 (1960); indeed, the defendant relied exclusively on his statistical analysis and challenged no particular aspect of the administration of the system.

It has become well-established that voter registration lists are appropriate for use in jury selection systems.21 Thus, there can be no objection to the fact that such lists were used here to determine how many persons should be taken from each community.22 There was no claim that the key men were selected in a discriminatory manner; indeed, the evidence adduced would belie such a claim. See n. 19, supra. The letter which the Commissioners sent to the key men — 291 F.Supp. at n. 7 — did not suggest that they confine their selections to persons of fair education, intelligence or esteem. Moreover, there was no reason to suspect that the key men would not be acquainted either personally or by reputation with a fair cross section of their local communities in Maine. Thus, we are satisfied that the persons recommended to the Commissioners as potential jurors were selected in a non-discriminatory manner.

There is, however, the question concerning the Commissioner’s exclusion of nearly half the persons recommended, in order to get down to the 900 persons needed for the jury pool. Part of that exclusion can be attributed to the fact that some key men sent too many names; when this occurred, the extra names were struck. Secondly, the Commissioner testified that some young adults who had been recommended were struck because they were out of the state for reasons relating to their education and were thus either ineligible under the uncontested qualifications for jury service — residence for one year — or were so far away that jury service would have posed a substantial hardship. Surely military service by some young men might put them in this same category. Thirdly, some young adults — particularly women, apparently — were excused from jury service when their questionnaires indicated that such service would be a hardship because of employment or the presence of young children at home. Finally, the Commissioner has indicated that almost 10 per cent of the recommended jurors were found unqualified for jury service for mental or physical reasons. Hearings on S. 383-387, S. 989, S. 1319 Before a Subcommittee of the Senate Committee on the Judiciary, 90th Cong., 1st Sess., at 964 (1967). See United States v. Caci, 401 F.2d 664, 671 (2d Cir. 1968), vacated on other grounds, 394 U.S. 310, 89 S.Ct. 1164 (1969).

Defendant has not contended that such individual excuses from jury service for hardship or disqualification were improperly granted. —see Thiel v. Southern Pacific Co., 328 U.S. at 224, 66 S.Ct. 974—it appears that *574members of the classes in question were excused only upon request or clear evidence from the questionnaire that excusable hardship did exist and not simply as a matter of course, That such exclusions were made by the Jury Commissioners rather than the district judge is not fatal. E. g., United States v. Caci, 401 F.2d at 671; United States v. Coppola, 296 F.Supp. 903, 904 (D.Conn. 1969).

We of course acknowledge that a system which persistently produced substantial and recognized underrepresentations of sociologically distinct groups would not be insulated from attack simply because it was fair on its face. Moreover, mere protestations or even evidence of subjective good faith would not dispel the inference that those who administer the system had purposed the results which they knew the system was producing, Here, however, the system was operated substantially in accordance with the 1960 Judicial Conference suggestions; the principle governing the selection of the several hundred key men was a neutral one; the letters from the Jury Commissioners to the key men were unexceptionable, as were the questionnaires sent to prospective jurors; there was no suggestion that the Commissioners applied improper standards in disqualifying or excusing individual jurors; the Commissioners made an effort to retain persons whose groups might otherwise be depleted;23 and, so far as we have reason to believe, this case — directed to the representation of groups of varying degrees of distinctness — is the first occasion that such protests have formally been brought to the attention of the Jury Commissioners.

We therefore conclude that this key man system was reasonably designed to obtain a fair cross section for the jury pools in southern Maine and that it has been administered in an effort to effectuate this design—Rabinowitz v. United States, 366 F.2d 34, 57 (5th Cir. 1966)—without discernible discrimination against any group.24 The Constitution’s mandate for a non-discriminatory jury selection system is not frustrated simply by the existence of certain inadvertent disparities arising from an otherwise fair system. As we said at the outset, it would be virtually impossible to assure a full representation of every cognizable group in every jury pool; all the Constitution requires in this regard is a jury selection system completely free of discrimination. This one was.

The sole remaining question is whether this case presents any need for us to exercise our supervisory power over the federal jury selection systems within our purview. Fay v. New York, 332 U.S. at 287, 67 S.Ct. 1613. We think not. As we observed in our discussion, there were some aspects of the system which could be improved to provide greater assurance of full representation for all cognizable groups. But Congress perceived these imperfections before they were brought to our attention; in 1968, Congress amended the existing federal jury system to provide that jury pools are to be chosen at random from a fair cross section of those eligible in the com*575munity. 28 U.S.C. § 1861 et seq. (1964). Accordingly, there is no need for us to offer any supervisory suggestions for improvements of the key man system.

Affirmed.