ROSS, Circuit Judge.

Robert Ross is a Missouri state prisoner serving consecutive sentences of 25 and 30 years for convictions, in Mississippi County, Missouri, of assault with intent to kill and first degree robbery. His petition for habe-as corpus, filed on March 17, 1976, is before this court for the second time. Ross contends, among other things, that the jury selection process used in his trial violated the due process and equal protection clauses of the fourteenth amendment.

Initially the district court denied the petition without a hearing. On appeal this court expressed concern over the racial composition of the jury panel and remanded for an evidentiary hearing. Ross v. Wyrick, 553 F.2d 51 (8th Cir. 1977). A magistrate held an evidentiary hearing, and on February 28, 1978, the district court entered an order again denying the requested relief. Ross appeals to this court.

The testimony before the magistrate revealed that Mississippi County utilized a “key man” system of jury selection. The master list from which jurors to serve at Ross’ trial were drawn was prepared in 1966 from individual voter registration cards. The county clerk read the names from the voter registration lists 1 to the jury commissioners. 2 As the clerk read the names the commissioners selected individuals they knew personally to make up the jury list. Of the 964 individuals selected for the master list, only 33 (3.4%) were black.

The parties stipulated that the 1970 population of Mississippi County was 16,647. Of that number 3,523 (21.16%) were black. Thus, the disparity between the black population and the number of blacks on the master list was' 17.7%.

When Ross’ case was called for trial on May 14, 1974, 23 of the 24 persons selected from the 1966 master list appeared in response to summons for jury duty. All were white. Ross moved to strike the jury panel but his motion was overruled. 3

Some of the venire panel were dismissed for cause and the judge ordered a special venire facias. Under the standard procedure used by the trial court at that time, the sheriff was free to select anyone he wanted to fill the six slots for the special venire facias. The sheriff testified that he chose four blacks and two whites because he was aware that the original panel included no blacks. Only three of the six persons so chosen were on the master jury list. The 12-person petit jury eventually chosen to try Ross included 2 blacks (16.6%), both of whom were on the master jury list.

In denying Ross’ petition for habeas corpus the district court found that the 1970 census showed that of the 7470 eligible jurors in Mississippi County, 1083 (14.5%) were black. 4 From this he found a disparity of *174 11.08% between the number of blacks eligible for jury service and the number on the master list. He found that although the figures suggest deliberate exclusion they are not alone conclusive. He also found no credible evidence of deliberate exclusion of blacks. He stated that although “[i]t may be necessary to concentrate on the selection process when one is tried before an all white jury,” that is not the case here. He concluded, “If error did exist in the master jury list, that error was corrected when the jury panel was selected.” We cannot agree.

When coupled with the obvious opportunity to discriminate present here in both the preparation of the master jury list and in the sheriff’s selection of additional jurors in a criminal case such as this, we find that either the 11.08% or the 17.7% disparity 5 is sufficient to sustain Ross’ prima facie case of jury discrimination. 6 See Turner v. Fouche, 396 U.S. 346, 360, 90 S.Ct. 532, 24 L.Ed.2d 567 (1970). See also Murrah v. Arkansas, 532 F.2d 105 (8th Cir. 1976).

There is no doubt that Mississippi County’s jury selection system presents the opportunity for discrimination. The Supreme Court has accepted the facial constitutionality of the “key man” selection system. See, e. g., Carter v. Jury Commission, 396 U.S. 320, 335-37, 90 S.Ct. 518, 24 L.Ed.2d 549 (1970); Akins v. Texas, 325 U.S. 398, 403, 65 S.Ct. 1276, 89 L.Ed. 1692 (1945); Smith v. Texas, 311 U.S. 128, 130-31, 61 S.Ct. 164, 85 L.Ed. 84 (1940). It has, however, noted that the system is susceptible of abuse. See, Castaneda v. Partida, 430 U.S. 482, 497, 97 S.Ct. 1272, 51 L.Ed.2d 498 (1977); Hernandez v. Texas, 347 U.S. 475, 479, 74 S.Ct. 667, 98 L.Ed. 866 (1954).

Here the commissioners did not select anyone who was not personally known by one of them. Furthermore, all blacks had a “C” following their names. Certainly the commissioners knew the race of every individual selected for the master list. Although the commissioners testified that they did not intentionally discriminate against any racial group, such general denials are not adequate to rebut a prima facie case of jury discrimination. See Alexander v. Louisiana, 405 U.S. 625, 632, 92 S.Ct. 1221, 31 L.Ed.2d 536 (1972); Turner v. Fouche, supra, 396 U.S. at 361, 90 S.Ct. 532.

What this court must decide, then, is whether the ultimate composition of the venire panel and the petit jury corrected *175 any constitutional deficiency in the jury selection process. We do not believe it did.

The state argues that the group of names from which the petit jury was actually drawn, the venire panel here, is the body that must represent the community. It cites Supreme Court cases allegedly supporting this contention. 7

What this argument overlooks is that here the master list from which the venire panels were drawn underrepresented blacks. The only reason the ultimate panel and petit jury had black members was because of the purely discretionary choice of the sheriff. 8 By keeping the number of blacks on the master lists disproportionately low, the state gained an advantage of having most prospective black jurors hand selected by the sheriff. The state cannot successfully contend that by allowing the chief law enforcement officer of the county to personally select the only black members on the venire panel in a criminal case, it overcame any constitutional deficiency in the selection process for the master list. Cf. Swain v. Alabama, 380 U.S. 202, 237, 85 S.Ct. 824, 13 L.Ed.2d 759 (1965) (Goldberg, J., dissenting).

We conclude that Ross has made a prima facie case of discrimination in the selection process of the jury which heard his case; that the state has not successfully rebutted that prima facie case nor shown that the constitutional deficiency was later corrected; and that the judgment of conviction is therefore unconstitutional and void.

Ross was chargéd in a valid two count information and is now legally detained pursuant to that information. He is entitled to be retried within a reasonable time. The district court is instructed to enter an order enabling the state to retry Ross within a specified period of time. If the state fails to try Ross within the specified time period the district court shall order the state to dismiss the information.

Reversed and remanded for proceedings consistent with this opinion.

Notes

. The Missouri Court of Appeals, in State v. Ross, 530 S.W.2d 457, 459 (Mo.App.1975) stated that “[t]he jury lists were drawn pursuant to § 494.240, RSMo 1969, from names taken from voter registration lists, and the names so selected did not indicate race.” The evidence adduced in this case indicates that the voter registration cards, the voter registration list prepared from those cards, and the master jury list picked from the voter registration list by the jury commission, had the letter (C) following each name of a black person.

. The circuit judge, the circuit clerk and the county’s three administrative judges comprised the jury commission.

. Robert Ross appealed to the Missouri Court of Appeals and his motion to transfer to the Supreme Court was denied. See State v. Ross, 530 S.W.2d 457 (Mo.App.1975). There is no question that he exhausted his state remedies and is properly before federal court.

. The district court must have taken judicial notice of these figures since they are not part of the record at the evidentiary hearing. The voter registration list for 1966 was admitted as evidence at the hearing. It indicated that of 8028 registrants, 1557 (19.4%) were black.

. In Turner v. Fouche, 396 U.S. 346, 360, 90 S.Ct. 532, 24 L.Ed.2d 567 (1970), the Supreme Court accepted petitioner’s showing of a 23% disparity between the percentage of Negro residents in the county as a whole and the percentage of Negroes on the jury list, coupled with the opportunity to discriminate in the selection process as establishing a prima facie case of jury discrimination.

In Castaneda v. Partida, 430 U.S. 482, 494, 97 S.Ct. 1272, 1280, 51 L.Ed.2d 498 (1977), the court stated that the second step in proving discrimination in a jury selection procedure is to prove the degree of underrepresentation by “comparing the proportion of the group in the total population to the proportion called to serve as grand jurors, * * (Emphasis added.)

In any event the state does not take a position that one figure rather than the other should be used.

. Many of the cases heretofore cited refer to the differences in the percentage of the underrepresented class living in the community vis-a-vis the percentage of that class on the master jury list. In this case the difference in those percentages was 17.7%. In Castaneda v. Partida, 430 U.S. 482, 486-87, 97 S.Ct. 1272, 51 L.Ed.2d 498 (1977), the average Mexican-American percentage of the total population of the county was 79.1%; the average percentage of Mexican-American grand jurors was 39%. This represented a 40% differential.

However, the facial comparison of the 17.7% differential in this case with the 40% differential in Castaneda is not truly indicative of the importance of the statistics. In this case the percentage of blacks on the master jury list was 3.4% compared to the 21.16% of black population in the county, or a ratio of one to six. In Castaneda the ratio of the percentage of Mexican-Americans as grand jurors to the percentage of Mexican-Americans in the population of the county, was only one to two.

The point is that in cases where the underrepresented class constitutes a large percentage of the population, a large percentage differential may not be as meaningful as a smaller percentage differential in a case in which the. underrepresented class constitutes a smaller percentage of the population.

. In Taylor v. Louisiana, 419 U.S. 522, 538, 95 S.Ct. 692, 702, 42 L.Ed.2d 690 (1975), the Supreme Court stated:

Defendants are not entitled to a jury of any particular composition, [citations omitted] but the jury wheels, pools of names, panels, or venires from which juries are drawn must not systematically exclude distinctive groups in the community and thereby fail to be reasonably representative thereof.

In Apodaca v. Oregon, 406 U.S. 404, 413, 92 S.Ct. 1628, 1634, 32 L.Ed.2d 184 (1972), the Court stated: All that the Constitution forbids, however, is systematic exclusion of identifiable segments of the community from jury panels and from the juries ultimately drawn from those panels; * * *.

. We do not reach the issue of whether a random drawing of a panel resulting in a fair representation of the community would correct any deficiency in the selection process for the master list. That did not happen here.