Madison County, Mississippi has elected its supervisors from the same five geographic districts since at least 1890. Shifts in population since that time have rendered the old districts seriously uneven in population. The local chapter of the NAACP and several private plaintiffs filed separate actions in 1983 seeking a reapportionment of the supervisor districts. The district court received proposed plans from the county, in which the NAACP joined, and from the private plaintiffs. That court rejected the county’s proposal and ordered that impending elections proceed under the plan suggested by the private plaintiffs. The county and the NAACP appealed, but the county withdrew after we declined to stay the district court’s order. Supervisory elections have now taken place under the new plan. We hold that the district court erred in rejecting the county’s plan in its entirety. Recognizing that resort to the county’s original plan may now be effectively foreclosed, we dissolve the district court’s injunction imposing the private plan and remand the case to the district court with directions.
I
Madison County lies just to the north of Jackson, Mississippi’s largest city. Like all Mississippi counties, Madison is divided into five supervisory districts, or “beats,” each of which elects a single member of the Board of Supervisors, which governs the county. Madison County has experienced explosive growth in the past few decades, especially at its southern end where the effect of Jackson’s expansion is niost directly felt. Despite its geographically uneven growth, the county has elected its five supervisors from the same five districts since at least 1890. By 1980, the population disparity had become extreme. The largest was district one, which included the whole of Canton, the county seat, and had a population of 16,963. The smallest was district five, which covered Madison’s rural northern end; its population was only 2,169. The total percentage-point variance from the mean was about 178%.1
Billy N. Cook and Robert Carter (the Cook plaintiffs) filed suit against the sitting supervisors and several Madison County election officials on June 6, 1983; the Canton branch of the NAACP filed a week later. Both complaints alleged that Madison’s supervisory districts violated the principle of one man, one vote announced by the Supreme Court in Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964), and both sought to enjoin the primary and run-off elections then scheduled for August 2 and 23, 1983. The NAACP also alleged that the existing district lines unconstitutionally diluted the political power of black voters in violation of the equal protection principle announced in early cases such as Fortson v. Dorsey, 379 U.S. 433, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965), Burns v. Richardson, 384 U.S. 73, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966), and White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973). See Rogers v. Lodge, *915458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982); City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47 (1980) (plurality); Jones v. City of Lubbock, 727 F.2d 364 (5th Cir.1984).2 The county defendants responded to both complaints by admitting that the existing supervisory districts were malapportioned under Reynolds v. Sims. They denied the NAACP’s claim that the district lines unconstitutionally diluted the voting strength of black citizens.
The district court consolidated the actions and on July 23, 1983, declared the population variance among the supervisory districts unconstitutional and enjoined the scheduled August elections.3 An evidentia-ry hearing was set for August 31, 1983, at which “all parties to this litigation and all other interested parties may come forward with plans” for redistricting Madison County.
All parties immediately set to work. The county retained an expert, Robert B. Hardy, and asked him to prepare a reapportionment plan. Hardy submitted his proposal at a public meeting August 16, 1983. For two hours the Board of Supervisors discussed the district lines proposed in Hardy’s draft plan with a crowd of citizens, consisting mostly of blacks concerned that their voting strength not be diluted. Several modifications were made, and the Board eventually approved the amended plan (the County Plan). The NAACP negotiated with the county in the formulation of its plan and participated in the public hearings; it joined in the county’s final version. The Cook plaintiffs also retained an expert, Joseph A. Lusteck, who prepared a second plan (the Cook Plan). Both the County Plan and the Cook Plan were submitted to the district court at the August 31 hearing.
The County Plan can only be understood in light of the existing district lines, for, as the district court recognized, the county proposed to retain the old lines whenever possible. Cook v. Luckett, 575 F.Supp. 485, 487 (S.D.Miss.1983). Madison County’s urban population was concentrated in two high-population districts: district one covered the north-central part of the county and included all of Canton, a city of about 11,000 people in 1980; district three comprised the southern section of the county and contained both Ridgeland and Madison, two towns just north of Jackson total-ling almost 8,000 people. The central problem in redistricting the county was to distribute these two urban pockets among the several supervisory districts and thereby to equalize the population of all five districts.
The County Plan achieved this population distribution by using the old districts as a starting point and creating corridors from outlying districts into Canton and down into the Jackson environs in southern Madison County. One corridor ran from sparsely populated district five in rural northern Madison County down along the west side of Highway 51 into Canton, where it broadened to take in the northern half of the city. A second corridor extended rural district four into a corner of district one and took in part of south Canton. The effect of these two population corridors was to leave district one bizarrely shaped: its western and most of its eastern sections were unchanged, but they were joined by only a narrow corridor that meandered through Canton, at times only a block or two wide. Similar corridors were used to distribute southern Madison County’s urban population among districts two and four. District two, in the far western part of the county, extended south and east to take in half of Ridgeland; district four in eastern Madison County expanded down the shore of the Ross Barnett Reservoir along a very narrow corridor to pick up urban population in *916the southern part of the county. These two corridors left district three a compact triangle shape including all of Madison and part of Ridgeway.
The Cook plaintiffs abandoned the old district lines entirely because they found them arbitrary and insufficiently tied to discernable geographical boundaries. Beginning afresh, Lusteck simply divided the county as a pie with Canton at its center. The Cook Plan described district lines along much more discernable boundaries than those proposed by the county and created districts shaped much more comprehensibly than the county’s “dumbbell” district one or its extremely narrow district four corridor.
Hardy testified that the County Plan created five districts of almost equal population, varying at the extremes only a total of 0.2% from the mean. That figure was challenged by the Cook plaintiffs’ expert, Lus-teck. Both he and Hardy used census data in drafting their plans, and both were forced to draw lines through the geographic units into which that data is organized. Lusteck had recalculated the population in split census units under the County Plan, and testified that the districts proposed by the county varied more from the mean than Hardy had figured. That error, together with an adjustment for population from an apartment complex in southern Madison County, created a total variance of 10.6% under Lusteck’s analysis of the County Plan. Lusteck calculated the total variance under the Cook Plan to be only 2.5%.
The two plans also differed in their effect on the voting strength of blacks in Madison County.4 The 1980 census revealed that 55.9% of Madison county’s population was black; the proportion of blacks to whites in the county had fallen steadily over the past 20 years from a 1960 level of 71.8%. Under the old apportionment, three districts contained populations over 70% black, one had a population of 56% black, and one had only 27.2% black. The County Plan would have created two districts roughly 70% black, one 63.9% black, and a fourth roughly even with 49.3% black. The Cook Plan created one district 76.8% black, one 68.3% black, one nearly even with 54.3% black, and two with about 40% black populations. The NAACP attacks the Cook Plan as diluting black voting strength.
Finally, the plans differ in that the county’s proposal was precleared by the Attorney General pursuant to section 5 of the Voting Rights Act of 1965, 42 U.S.C. § 1973c (1976). The county submitted its plan to the Attorney General on September 8, 1983, just after the evidentiary hearing. The submission was given expedited consideration, and on September 22, 1983, the Attorney General informed the county that he would not object to the County Plan, although he reserved the right to reconsider his decision should new information be submitted within the remainder of the sixty-day period. The Cook plaintiffs never submitted their plan for preclearance under section 5.
The district court rendered its decision on October 12, 1983. 575 F.Supp. 485. It rejected the County Plan, finding that the plan’s districts were “contorted” in shape and that they “fail[ed] on their face to take communities of interest into account.” Accepting Lusteck’s testimony, the court declared the County Plan unconstitutional because the county had not offered proof sufficient to justify a total variance of 10.6%. The court rejected several explanations offered by the county, including the *917desire to equalize the road mileage within each district. It then stated that the County Plan used “only a very few of the existing district lines.”
However, because the existing districts are so outdated they are a poor basis on which to build the new districts. Madison County has waited for almost one hundred years for new districts; they need not be so poorly designed. Although the Board-NAACP paid lip service to the use of existing lines, this Court finds that this was not a bona fide goal of the plan.
575 F.Supp. at 490-91.
The district court chose instead to adopt the Cook Plan, which it considered a “rational approach to dividing the county.” The total variance under the Cook Plan was only 2.5%, and the court found the plan to reflect established communities of interest, to take advantage of rational, visible boundaries, and not significantly to dilute black voting strength. Although it adjusted. a few precinct lines, the district court adopted the supervisory district lines proposed in the Cook Plan without change. Finally, the court imposed a timetable for conforming the voter rolls to the new districts and making other preparations for special supervisory elections, to be held in January 1984.
The county and the NAACP immediately appealed and moved the district court to stay the effect of its injunction pending appeal. Fed.R.App.P. 8. Applying the standard we expressed in Ruiz v. Estelle, 666 F.2d 854, 856 (5th Cir.1982), cert. denied, 460 U.S. 1042, 103 S.Ct. 1438, 75 L.Ed.2d 795 (1983), and Ruiz v. Estelle, 650 F.2d 555, 565 (5th Cir.1981), cert. denied, 460 U.S. 1042, 103 S.Ct. 1438, 75 L.Ed.2d 795 (1983), the district court found that the appeal presented substantial legal questions, but that the “equities” balanced against a further delay in elections. Thus, it found that neither the county nor the NAACP would be irreparably harmed by elections under the Cook Plan; that the Cook plaintiffs would be substantially harmed by a delay in the elections; and that the public interest would be disserved by allowing an unconstitutionally elected Board of Supervisors to govern Madison County while the constitutional plan imposed by the district court was challenged on appeal “on technical grounds.” The county then moved this court to stay the district court’s order pending appeal; we denied the motion on December 14, 1983. Finding special elections effectively unavoidable, the county dismissed its appeal; the NAACP remains as the sole appellant.5
II
We begin with the NAACP’s argument that the district court erred in its wholesale rejection of the County Plan. That plan, contends the appellant, represented the considered judgment of those elected to represent the citizens of Madison County, and it was entitled to respect. The NAACP relies principally on Upham v. Seamon, 456 U.S. 37, 102 S.Ct. 1518, 71 L.Ed.2d 725 (1982) (per curiam), in arguing that the district court’s response to the County Plan should have been limited to the plan’s constitutional flaws, and that the only intrusion into legislative judgment justified in this case was some slight readjustment of boundaries as necessary to bring the variance among the districts to within the constitutional tolerance. We must agree.
Upham involved the recent efforts to reapportion Texas’ congressional districts. The state legislature enacted a plan, SB1, and submitted it to the Attorney General for preclearance. Before the Attorney General could act, suit was filed challenging SBl’s validity under both the Constitution and section 2 of the Voting Rights Act, 42 U.S.C.A. § 1973 (West Supp.1984). The Attorney General then responded to the Texas submission, objecting to the lines that defined two districts in south Texas. Faced with an unenforceable SB1 and a previous apportionment that clearly violat*918ed the Constitution, the three-judge district court resolved the Attorney General’s objections to the two south Texas districts and imposed SB1 as a court plan with that change and one other: the court redrew the district lines in the Dallas area without expressly finding those lines to violate either the Constitution or the Voting Rights Act.6
The Supreme Court summarily reversed. SB1 had failed preclearance and could not be effectuated in the enacted form. McDaniel v. Sanchez, 452 U.S. 130, 101 S.Ct. 2224, 68 L.Ed.2d 724 (1981). The district court therefore faced the task of imposing an interim apportionment plan to allow elections to go forward, and in doing so was required to “reconcil[e] the requirements of the Constitution with the goals of state political policy.” Upham, 456 U.S. at 43, 102 S.Ct. at 1522 (quoting Connor v. Finch, 431 U.S. 407, 414, 97 S.Ct. 1828, 1833, 52 L.Ed.2d 465 (1977)). The Court held that “[a]n appropriate reconciliation of these two goals can only be achieved if the District Court’s modifications of a state plan are limited to those necessary to cure any constitutional or statutory defect.” Id. at 43, 102 S.Ct. at 1522.
Just as the district court in Up-ham was bound to defer to the political judgment embodied in a state’s plan of congressional apportionment, so was this district court bound to defer to the choices made by Madison County in formulating its apportionment scheme.7 See Wise v. Lipscomb, 437 U.S. 535, 539-40, 98 S.Ct. 2493, 2497, 57 L.Ed.2d 411 (1978) (apportionment of city council); id. at 549, 98 S.Ct. at 2501 (Powell, J., concurring); Wyche v. Madison Parish Police Jury, 635 F.2d 1151, 1158 (5th Cir.1981). Like the Upham Court, we recognize the several cases imposing on district court a heightened obligation when formulating an interim remedial plan: such court-imposed plans must avoid the use of multi-member districts and must achieve a greater degree of population equality than is required of legislative plans. See Wise v. Lipscomb, 437 U.S. at 540, 98 S.Ct. at 2497; Connor v. Finch, 431 U.S. at 414, 97 S.Ct. at 1833. That heightened standard will apply, of course, when a district court is forced by an uncorrectably invalid legislative plan to impose a scheme submitted by a private litigant or to formulate its own plan. Even then the federal court must honor state policies to the greatest extent possible when choosing among available plans or fashioning its own. But when the court is presented with a legislative plan whose constitutional or statutory flaws are capable of correction by minor adjustments, Upham requires the court to minimize violence to those legislative policies embodied in the plan by changing it only to the extent necessary to cure its cognizable flaws. Upham, 456 U.S. at 41-3, 102 S.Ct. at 1521-22. See also White v. Weiser, 412 U.S. 783, 795, 93 S.Ct. 2348, 2354, 37 L.Ed.2d 335 (1973).
The facts of this case serve well to explain why federal courts should and must defer to legislative plans of apportionment. While the maps depicting its result may seem odd, Madison County’s political process involved just the sort of give-and-*919take between citizens and their elected officials that federal courts are unable to achieve. Unless a showing is made to render that give-and-take somehow suspect, we must acknowledge that process as the proper means toward the essentially political end of reapportionment. As Judge Wisdom has written, “[t]he least representative branch of the government must take care when it reforms the most representative branch.” Marshall v. Edwards, 582 F.2d 927, 934 (5th Cir.1978), cert. denied, 442 U.S. 909, 99 S.Ct. 2820, 61 L.Ed.2d 274 (1979). We therefore act with the deference due legislative plans in requiring that, when a district court is presented with a legislative apportionment scheme that violates constitutional or statutory norms, it may reject that scheme only to the extent necessary to correct the specific deficiency found by the district court to exist.
A
In rejecting the County Plan, the district court relied expressly on the population variance among the supervisory districts proposed by the county. The Supreme Court has established that legislatively enacted apportionment plans with districts that vary from the mean less than a total of 10% present only “minor deviations” insufficient to make out a prima facie case of invidious discrimination under the Fourteenth Amendment. E.g., Connor v. Finch, 431 U.S. 407, 418, 97 S.Ct. 1828, 1835, 52 L.Ed.2d 465 (1978). See Brown v. Thomson, 462 U.S. 835, —, 103 S.Ct. 2690, 2696, 77 L.Ed.2d 214 (1983). Cf. Abate v. Mundt, 403 U.S. 182, 185, 91 S.Ct. 1904, 1907, 29 L.Ed.2d 399 (1971) (slightly larger deviations permissible in local apportionment plans). Madison County’s plan exceeded this level of “minor deviation” by less than one percentage point. Even assuming that the district court correctly concluded that the 10.6% variation unjustifiably exceeded constitutional limits, it was empowered by that finding only to return the county’s planners to their drawing boards or itself to amend the County Plan sufficiently to achieve an acceptable population equality among the districts.
The County Plan’s 10.6% total variance was caused by two factors. First, the trial court credited the Cook plaintiffs’ expert, Lusteck, when he testified that several districts proposed by the county varied more widely from the mean than was calculated by Hardy, the county’s expert. Both men counted population on the basis of the geographic units into which the data from the 1980 census is organized. Their difference was caused by the methods they each used to evaluate the numerical effect of splitting these census units when drawing district lines. We are bound, of course, by the district court’s credibility choice. Nevertheless, this error of the county explains only about half of the total variance held by the district court to exceed constitutional levels. The county’s proposed district three was thought by the county’s expert to vary from the mean by less than one percent, but Lusteck testified that its population fell below the mean by 4.9%. Nothing in this record explains why this could not easily have been narrowed further by minor adjustments in the county’s proposed district lines.
The remainder of the 10.6% total variance resulted in major part from the dis-tricting of Hickory Knoll Apartments in southern Madison County. The county had included Hickory Knoll along with an adjacent complex in its district two by slightly diverting the line between that district and district four. Lusteck had understood the County Plan to place Hickory Knoll in district four; when he recalculated his corrected figures under the County Plan to include 260 additional residents in district two, that district’s population rose to exceed the mean by 5.7%. Added to district three’s 4.9% negative variance, the increased variance of district two brought the total to 10.6%. Had the complex been counted in district four, as Lusteck originally understood, neither district two nor district four would have exceeded the mean by more than 3.2%; even without any adjustment in district three’s variance, the *920total would have fallen to 8.1%, well within constitutional levels for a legislative plan.8
We do not suggest that this adjustment would have been wise or that it was the only change that might have reduced the total variance under the county’s proposal to within constitutional limits. It is apparent, however, that the relatively slight unevenness in population created under the County Plan as Lusteck computed it could easily have been corrected by small adjustments in the county’s proposed district lines. The district court went far beyond such minor adjustments, rejecting the County Plan in its entirety, and we are bound by Upham to disapprove that decision.
B
The district court also rejected the County Plan because its districts were bizarrely shaped: “Madison County has waited for almost one hundred years for new districts; they need not be so poorly designed.” 575 F.Supp. at 491. We agree that the county’s proposed district lines were more than just odd. Indeed, they seem to us, as they did to the court below, to respond poorly to commonly understood policies that govern apportionment planning. But after Upham, questions of policy are reserved for legislative resolution. The district court was authorized to alter the county’s legislative plan only in those ways necessary to remedy a constitutional or statutory vice. Because this record does not support the conclusion that the districts were constitutionally flawed by their bizarre shapes, we hold that the district court erred in rejecting the County Plan on this basis.
The Supreme Court has never applied the Fourteenth Amendment to reject a legislative plan of apportionment solely because the plan proposed weirdly shaped districts, and its opinions have only once even suggested a unified constitutional analysis for reviewing the shapes of voting districts. Karcher v. Daggett, 462 U.S. 725, —, 103 S.Ct. 2653, 2667, 77 L.Ed.2d 133 (1983) (Stevens, J., concurring); id. at -, 103 S.Ct. at 2687 (Powell, J., dissenting). Cf. Reynolds v. Sims, 377 U.S. 533, 578-79, 84 S.Ct. 1362, 1390, 12 L.Ed.2d 506 (1964) (“Indiscriminate districting, without any regard for political subdivision or natural or historical boundary lines, may be little more than an open invitation to partisan gerrymandering”); Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960) (racial gerrymander invalidated under Fifteenth Amendment). Karcher presented the Court with New Jersey’s remarkable effort at congressional reapportionment after the 1980 census. The court held that the legislative plan’s total variance of less than one percent violated the specially exacting standard applied under Article I, § 2 to congressional apportionments.9 Justice Stevens concurred, suggesting as an alternative ground for rejecting New Jersey’s plan the bizarre configuration of its proposed congressional districts. Justice Powell dissented from the majority’s view that New Jersey’s plan was invalid because of its population variance, but agreed in general with Justice Stevens’ suggestion that the contorted district lines proposed by the state revealed an unconstitutional political gerrymander. Assuming for the purposes of this case alone that the analyses suggested by Justices Stevens and Powell are valid, we conclude that the Cook plaintiffs made an inadequate showing to shift to the county the task of justifying the shapes and locations of the districts it proposed.
The claim in this case is a denial of equal protection. Both the Cook plaintiffs and the NAACP alleged that Madison County’s old supervisory districts diminished the voting power of those in overpopu*921lated districts and enhanced the power of those in underpopulated districts. Reynolds v. Sims established twenty years ago that the force of votes cast for members of an area-wide governing body must be roughly equal within that area, and district-ing schemes that disenfranchise one group in favor of another are therefore invalid under the Equal Protection Clause. Appor-tionments that work no selective disenfranchisement but are merely unwise do not violate the equal protection standards that have developed since Reynolds v. Sims, and their perceived lack of wisdom is not to be corrected in the federal courts.
Justice Stevens’ rationale for reviewing the shapes of voting districts is tied closely to the equal protection roots of Reynolds v. Sims. His is a theory of political gerrymandering in which one challenging the contorted shapes of voting districts must prove first that they belong to a “politically salient class” of voters whose “geographical distribution is sufficiently ascertainable that it could have been taken into account in drawing district boundaries,” and second “that their proportionate voting influence has been adversely affected by the challenged scheme.” Karcher, 462 U.S. at —, 103 S.Ct. at 2672 (Stevens, J., concurring). If a group of plaintiffs can bring themselves within the ambit of the Equal Protection Clause by making these two showings, then they may present their prima facie case of discrimination by showing, for example, that the districts into which they are grouped are so “dramatically irregular” as to call for an explanation. Id. Only then might the governmental proponent of the plan be asked to justify its district lines according to neutral criteria.
The Cook plaintiffs have failed to bring themselves within any “politically salient” group of voters and have therefore failed to trigger the protections of the Fourteenth Amendment. This is simply not a case in which persons of a particular ethnic or political group face selective disenfranchisement under the County Plan. Without that element of selectivity, the Equal Protection Clause does not forbid unwise districting schemes.10
We therefore hold that neither the small variation in population among the county’s proposed districts nor the rather bizarre shapes of several of those districts justified the district court in rejecting the plan and adopting in its stead the completely different districting scheme proposed in the Cook Plan. We now turn to the difficult problem created by the recent elections.
Ill
Two related circumstances complicate our disposition of this case. First, the original county defendants, including the members of the sitting Board of Supervisors, dismissed their appeal after their applications to stay the district court’s order were denied. Second, the district court’s undisturbed order imposing the Cook Plan has been followed by elections under that plan, substantially changing the membership of the board. We decline to upset the recent elections, but we remand the case with directions to the district court designed to produce a permanent plan for apportioning Madison County’s supervisory districts consistently with the dictates of the Fourteenth Amendment and the Voting Rights Act.
We need not pause long in deciding to leave the recent elections untouched. First, we are unsure whether the NAACP urges us to upset the elections. Second, our decisions reveal a strong reluctance to undertake the “drastic, if not staggering” remedy of voiding a local election. Bell v. *922Southwell, 376 F.2d 659, 662 (5th Cir.1967). We reviewed those cases in Saxon v. Fielding, 614 F.2d 78 (5th Cir.1980), and concluded that elections will not be set aside absent “serious voting violations or aggravating factors, such as racial discrimination or fraudulent conduct____” Id. at 79. The January elections in Madison County took place in districts entirely consistent with the Fourteenth Amendment’s quantitative requirement of one man, one vote. With regard to its qualitative requirement, we need only conclude that the Cook Plan’s districts did not so seriously dilute the political strength of Madison County’s black voters that we would consider setting aside the elections on that ground.11 See United States v. Louisville Municipal Separate School District, 557 F.Supp. 1168, 1172 (N.D.Miss.1983) (three-judge court). We so conclude. Thus, nothing of substance leads us to void the elections.
Finally, we need face only obliquely the knotty problem of preclearance presented by the lower court’s imposition of a private party’s unprecleared redistricting proposal. Having determined on full review that the district court erred in rejecting the County Plan, we will not expound in dictum on this second possible ground for rejecting its imposition of the Cook Plan. In declining to disturb the elections, we recognize the prevailing uncertainty whether the Voting Rights Act requires preclearance of remedial plans submitted by private parties in reapportionment litigation. Compare 42 U.S.C. § 1973c (1976) (requiring preclearance whenever a covered “state or political subdivision ... shall enact or seek to administer” a new districting plan) with S.Rep. No. 295, 94th Cong., 1st Sess. 18-19, reprinted in 1975 U.S.Code Cong. & Ad. News 774, 784-85 (preclearance required whenever any litigant proposes plan); cf. McDaniel v. Sanchez, 452 U.S. 130, 101 S.Ct. 2224, 68 L.Ed.2d 724 (1981) (legislative plan must be precleared before submission to district court). The Supreme Court long ago declined to set aside elections for violations of the Voting Rights Act not clearly established at the time of the election. See Allen v. State Board of Elections, 393 U.S. 544, 571-72, 89 S.Ct. 817, 835, 22 L.Ed.2d 1 (1969). For these reasons, we leave untouched the practical and immediate effect of the district court’s order imposing the Cook Plan: we will not disturb the elections.
The future application of the court-imposed Cook Plan is far less certain. We have already noted that the business of reapportionment is “a legislative task which the federal courts should make every effort not to preempt.” Wise v. Lipscomb, 437 U.S. 535, 540, 98 S.Ct. 2493, 2497, 57 L.Ed.2d 411 (1978); Connor v. Finch, 431 U.S. 407, 414-15, 97 S.Ct. 1828, 1833-34, 52 L.Ed.2d 465 (1977). When the district court correctly found Madison County’s existing apportionment to violate the one person, one vote principle of Reynolds v. Sims, it was bound to afford the county an opportunity to develop a new plan of apportionment, see Wise v. Lipscomb, 437 U.S. at 539-40, 98 S.Ct. at 2497; Ramos v. Koebig, 638 F.2d 838, 843 (5th Cir.1981), and to adjust the county’s proposed plan only to the extent necessary to correct any constitutional or statutory flaws. Upham v. Seamon, 456 U.S. 37, 102 S.Ct. 1518, 71 L.Ed.2d 725 (1982) (per curiam). The district court afforded Madison County an opportunity to propose a plan, but its wholesale rejection of that plan — however well intended — went far beyond any remediable flaws in the plan.
The record available to us discloses that the district court did not limit the terms of its injunction to impose the Cook Plan as only a temporary districting scheme in order to enable the delayed elections to proceed. Many federal courts, forced by the perceived invalidity of a legislative plan to alter that plan or to impose a plan of their own, have put the remedial alteration or plan in place only temporarily so that elec*923tions may occur. E.g., Ramos v. Koebig, 638 F.2d at 838; Terrazas v. Clements, 537 F.Supp. 514, 527-28 (N.D.Tex.1982) (three-judge court).
In light of our conclusion that the district court erred in rejecting the County Plan outright, we will dissolve such part of the lower court’s injunction as may now remain in force and remand the case to the district court. That court is directed to invite Madison County to submit a duly enacted and precleared legislative plan of apportionment. We recognize that those who now govern in Madison County may no longer support the original County Plan. Indeed, it seems likely they may choose to propose a plan substantially similar to the Cook Plan under which they were elected. That is their right as elected officials. But if the districts set out in the Cook Plan and imposed by the order we vacate today are to become permanent, it will be because Madison County chooses that course. In so choosing, it would adopt a districting plan subject to section 5’s preclearance requirement. If the county’s proposed and pre-cleared plan meets constitutional requirements, the district court will approve the plan as establishing Madison County’s permanent supervisory districts.
VACATED and REMANDED.
. With Madison County’s total 1980 population of 41,613 and five districts, the perfect district would have a population of 8323. Under the old districting scheme, district one exceeded that mean by 103.8% and district five fell short by 73.9%. We have calculated that the old districts varied from the mean an average of 73%.
. Both plaintiffs also claimed that an impending purge of the county's voter rolls would adversely affect the voting strength of minorities. That claim has apparently been overshadowed by recent events. In any event, it is not before us.
. The district court’s later memorandum opinion setting forth reasons for enjoining the elections reveals that he actually faced challenges to the supervisory districts in five Mississippi counties. Cook v. Luckett, 575 F.Supp. 479 (S.D.Miss.1983). We have the luxury of considering only Madison County in this appeal.
. Remarkably, this record contains no population data reflecting the racial breakdown of Madison County's registered voters. We are therefore forced to use raw population figures in relating the racial percentages in Madison County’s existing and proposed supervisory districts. The Cook plaintiffs apparently made some effort to adjust raw population data to reflect the number of citizens of voting age, but the record affords us almost no information about the mechanism by which Lusteck managed even that adjustment. See 3 Record 104-OS (testimony of Lusteck); Cook exhibit 5. Our consideration of the "qualitative” aspect of voting cases — dilution of group voting strength— can be only as exacting as the record before us allows. In light of the posture of this case and of our holding today, we need not remand for a fuller record.
. We are told that the special elections took place in January 1984, and that they resulted in the election of the first two black supervisors in Madison County’s history.
. Seamon v. Upham, 536 F.Supp. 931 (E.D.Tex.), rev’d in part, 456 U.S. 37, 102 S.Ct. 1518, 71 L.Ed.2d 725 (1982). Judges Johnson and Justice joined in rejecting the Dallas districts, but only Judge Justice considered them unconstitutional. Judge Johnson wrote that once any constitutional or statutory flaw is found in a legislative plan, the court thereafter must impose a plan of its own and must design that plan according to the special standards applicable to court-imposed apportionments. 536 F.Supp. at 939-50. Although the Dallas districts would apparently not have been unconstitutional if enacted by Texas, they did not meet the standards applicable to court plans. Judge Parker dissented, expressing the view that the Dallas districts enacted in SB1 were to be followed unless unconstitutional. Id. at 1028.
. The district court expressed uncertainty whether the County Plan was a legislative proposal entitled to deference, suggesting that the NAACP’s participation may have undermined that status. 575 F.Supp. at 491. There are no facts in the record to suggest why the County Plan should not be deemed legislative in character. The district court’s opinion reflects its assumption that the plan was legislative, id. So shall ours.
. Counting Hickory Knoll’s 260 residents in district two sent that district’s population from 8538, or 2.6% over the mean, to 8798, or 5.7% above the mean; counting it in district four would send that district's population from 8331, or .1% over the mean, to 8591, or 3.2% above the mean.
. Four justices dissented on the theory that so small a deviation was constitutionally insignificant. 462 U.S. at —, 103 S.Ct. at 2678 (White, J., dissenting).
. We repeat our conclusion in Jones v. City of Lubbock, 727 F.2d 364, 370 n. 2 (1984), that what may once have been separate "dilution” and "gerrymander” lines of analysis have now merged. The challenge mounted in this case to the shapes of the county's proposed districts fails at the first step, for no matter whether the claim is analyzed according to the gerrymander cases such as Gomillion or dilution cases like Kirksey v. Board of Supervisors, 554 F.2d 139 (5th Cir.) (en banc), cert. denied, 434 U.S. 968, 98 S.Ct. 512, 54 L.Ed.2d 454 (1977), the Cook plaintiffs failed to identify any class of voters whose ballots were weakened by the lines proposed by the county.
. Because of our holding and disposition in this case, we need only consider the question of dilution in deciding whether to overturn the elections. We have therefore not found it necessary to remand for development of more adequate figures on the rates of registration among black and white voters in Madison County.