The reapportionment of legislative bodies to accord one person one vote assures equal protection of voters as electors. Even when voters are afforded numerical equality as members of the polity, racial discrimination may prompt a subtle form of gerrymandering that in effect makes a vote by the voter of one race less important than the ballot of the voter of another race, and thus violates the fourteenth amendment by denying one group of voters equal protection because of their race even where their ballots have the same apparent weight. Yet it is fundamental that the Constitution does not require proportional representation of voters by *1155race or any other characteristic. We here consider both the extent to which, and the manner in which, the race of the population of a Louisiana parish can or must be considered by a court that is reapportioning the parish’s governmental bodies.
I.
Madison Parish is located on the Mississippi River in northeast Louisiana. The Police Jury is its parochial governing body. The School Board has the usual responsibilities implied by its name. These two public bodies are elected by the voters of the Parish. According to a special census conducted before the 1980 decennial census, it has a total population of 14,472, 58% of whom are black. The parish is essentially rural but 66% of the total population lives in the town of Tallulah or its immediate environs. This town area is heavily black, while the more sparsely populated rural area is preponderantly white.
For almost a century prior to 1963 no black person had been registered to vote in the parish. In that year, a federal court order required the parish officials to permit black persons to register by applying to them the same qualifications used for white applicants. United States v. Ward, 222 F.Supp. 617 (W.D.La.1963), rev’d, 349 F.2d 795 (5th Cir. 1965), modified on rehearing, 352 F.2d 329 (5th Cir. 1965). Thereafter several elections for municipal and parochial offices were set aside and new elections ordered on the ground that black voters and candidates had been illegally deprived of their right to exercise the franchise. See Brown v. Post, 279 F.Supp. 60 (W.D.La.1968); United States v. Post, 297 F.Supp. 46 (W.D.La.1969); Toney v. White, 348 F.Supp. 188 (W.D.La.1972), aff’d in part and rev’d in part, 476 F.2d 203, modified on rehearing, 488 F.2d 310 (5th Cir. 1973) (en banc). See generally E. Kaplan & J. Stanzler, Voting Rights: A Case Study of Madison Parish, Louisiana, 38 U.Chi.L.Rev. 726 (1971). Nonetheless, no black person was elected to parochial office until 1971.
In July, 1971, the parish was reapportioned under a court order. The districting plan required the creation of four wards from which eight persons would be elected to the Police Jury and eight to the School Board. Three were multi-member districts and one was a single-member district. In elections held pursuant to that plan, three black persons were elected to the Police Jury and three to the School Board.
After the court drew its plan and the 1971 elections were held, two low income housing projects, Madison Community Apartments and Starr Lodge Apartments, were built outside of the town limits of Tallulah. More than 700 persons, about five percent of the total parish population, live in these projects. In July, 1972, the town of Tallulah annexed the area in which the projects are located. The extension of the town boundaries was treated by the registrar of voters as an extension of the boundary lines of Ward I of the parish districting plan. Thus, while the two low income projects were physically located in what had been a predominantly rural ward, the residents of those projects and the surrounding area were registered to vote and in fact voted in a town ward. This de facto change of the district lines was made without either court approval or submission to the United States Department of Justice as required by Section 5 of the Voting Rights Act of 1965, 42 U.S.C. § 1973c. The result was to increase the population of the town ward in relation to the other districts, thus creating underrepresentation for the residents of that ward on the Police Jury and School Board. The plaintiffs, therefore, petitioned the federal court for a new redistricting.
Thé court appointed a qualified demographer, Kenneth Selle, to make a parish census and, after consulting with the two parochial bodies and their counsel, to draft a plan of reapportionment composed of single-member districts. In the eight year interim, the total parish population had declined slightly, and the percentage of black population had fallen from 60% in 1970 to 58% in 1978. Mr. Selle prepared three proposed plans, one of which, Plan 2, is not urged by either party. The first plan (Plan *11561) proposed eight districts. Two are rural, lying east and west of Tallulah. The remaining six each include portions of Tallu-lah and the area immediately surrounding the town. A more complete description of the proposal is set forth in the footnote.1 Its essential features are that four districts, the two rural districts and two of the town districts, have white population majorities, while the population of the four remaining town districts is almost exclusively black. This was a preliminary plan and Mr. Selle never refined it. He testified, however, that, with slight modification, the deviation factor (plus 8.4% added to minus 3.3%, a total of 11.7%) could be substantially reduced. As the population figures indicate, in the event of block voting along racial lines, Plan 1 would result in the election of four black and four white representatives to each public body.
Plan 3, recommended to the court by Mr. Selle and endorsed by the Police Jury and School Board, also proposes eight districts. It differs from Plan 1 chiefly in that the rural area of the parish is divided into three districts instead of two, leaving five town districts. The details of this plan are set forth in the footnote.2 The plan suggests five districts with black majority population. However, district 1 would have a black population of 53% and district 3 would have a black population of 62% while in Plan 1, no majority black district would have less than 93% black population. Therefore, it would be easier for a white candidate to be elected in the two districts in Plan 3 having only 53% and 62% black population. The deviation factor of this plan is 4.11%.
Plan 3 represents a conscious effort by the demographer to devise a plan containing three rural districts. To accomplish this goal and conform to one person, one vote standards, Selle had to “borrow” portions of the northern area of Tallulah to increase the populations of districts 1 and 3, bringing the number of persons in those districts closer to the median number. Those town areas include the public housing projects and are all black in population.
Mr. Selle testified that, in his opinion, the two plans are essentially equal and that either plan would satisfy the basic criteria for constitutional adequacy. He recommended Plan 3 because Madison Parish is an agricultural area and its economy is based on agriculture. The governing bodies must concern themselves with activities that affect the agricultural areas. Therefore, Plan 3 considers the orientation and interests of the parochial population.
The plaintiffs contend that, while Plan 3 proposes five districts with a majority black population, black candidates could be elected from only three of those districts because the black population figures in districts 1 (53%) and 3 (62%) do not reflect voting strength. They urge this court to adopt a rule that, to remedy demonstrated discrimination against black voters a court must adopt the plan most likely to result in proportionate election of black representatives.
To buttress their objection to Plan 3, the plaintiffs sought to introduce evidence to prove that this plan would in fact result in the election of only three black persons, but the Court ruled the evidence inadmissible. *1157In an offer of proof, the plaintiffs took the testimony of a black police juror who is also the manager of one of the low income housing projects on the northern edge of Tallu-lah. He testified that, if Plan 3 were adopted, the apartments and his residence would be in district 3, and there were not enough black residents of voting age in the district to form even a potential black majority of voters. This testimony was confirmed by voter registration statistics which showed that 59% of the registered voters in district 3 as of December 31, 1978, were white while only 41% of those registered were black.3 The excluded voter registration data also showed that 8,590 voters are registered in the parish and 52% of these are white. Thus, while the parish population is 58% black, only 48% of the registered voters are black.
Having excluded evidence of the plan’s effect at the polls, the district judge followed the demographer’s recommendation and adopted Plan 3, stating that he did so because Plan 1 would tend to dilute the voting strength of the rural areas of the parish. In addition, the district judge noted that Plan 1, even had it been modified as proposed by Mr. Selle, would have a relatively high deviation factor.4
Thereafter, the district judge who had adopted Plan 3 retired. The Police Jury election was scheduled for 1979. An appeal having been taken without a stay order, the successor district judge ordered the election conducted pursuant to Plan 3. That election was held, and the members of the Police Jury were elected to hold office for four years. The next election is scheduled for the fall of 1983, with the new Police Jury to take office in 1984.
Pursuant to state statute, the School Board members had been elected for staggered terms. In 1980, the Louisiana legislature amended the statute to require concurrent terms. 1980 La.Acts, No. 285, amending La.R.S. 17:52 (1950). To implement the change without undue disruption, the legislature provided that school board members elected in 1980, 1982 and 1984 shall serve six, four and two year terms respectively. All of the seats will be filled in the 1986 election and members will thereafter serve concurrent terms. Under the district court order mandating the use of Plan 3 for School Board elections, two of the eight School Board members were elected in 1980 for six year terms. The district court’s order extended the terms of three incumbents to 1982 and the terms of the remaining three incumbents to 1984. Thus, three School Board members will be elected in 1982 for four year terms and, in 1984, three members will be elected for two year terms.
Of course, one other event, of which we may take judicial notice has occurred: the 1980 census has been conducted, and its results are being released on a county by county basis as they are completed. Although total population figures for Madison Parish have recently been made available, other demographic data, such as age and race, have not yet been released. No census data for Madison Parish is, as yet, officially published. Louisiana requires its parochial bodies to examine their apportionment within six months after the official release of every decennial census to determine if there exists any substantial variation in the representation of the election districts. After this examination, the body must by ordinance either declare its apportionment to be equitable and continue its existing plan or provide for a new apportionment plan to be made effective at the end of the term of incumbent officials. La. R.S. 33:1411 (Supp.1968).
II.
The Supreme Court entered the “political thicket” of legislative apportionment in Baker v. Carr, 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962), by holding justiciable a *1158challenge to the constitutional validity of an apportionment scheme under the fourteenth amendment’s equal protection guarantee. In Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964), the Court held unconstitutional the election of representatives to a state legislature from districts substantially unequal in population because this process dilutes the vote of members of more populous districts and thus denies them equal protection.
Thereafter, the Court applied the one person, one vote precept to reapportionment of the voting constituencies in units of local government. Abate v. Mundt, 403 U.S. 182, 91 S.Ct. 1904, 29 L.Ed.2d 399 (1971); Hadley v. Junior College District, 397 U.S. 50, 90 S.Ct. 791, 25 L.Ed.2d 45 (1970); Avery v. Midland County, 390 U.S. 474, 88 S.Ct. 1114, 20 L.Ed.2d 45 (1968). We have applied the principle to local government units, such as school boards, Panior v. Iberville Parish School Board, 498 F.2d 1232 (5th Cir. 1974); counties, Kirksey v. Board of Supervisors of Hinds County, 554 F.2d 139 (5th Cir.) (en banc), cert. denied, 434 U.S. 968, 98 S.Ct. 512, 54 L.Ed.2d 454 (1977); and other parochial bodies, Dundee v. Orleans Parish Board of Supervisors of Elections, 434 F.2d 135 (5th Cir. 1970), vacated and remanded on other grounds, 403 U.S. 915, 91 S.Ct. 2231, 29 L.Ed.2d 692 (1971).
The one person, one vote principle commands that constituencies include approximately equal numbers of voters, so that the weight of individual votes in larger districts will not be substantially diluted and individuals in those districts will not be deprived of fair and effective representation. Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964); White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973). The population equality principle requires a quantitative analysis typically focusing on whether deviations from the average or median population district are impermissibly large. Gaffney v. Cummings, 412 U.S. 735, 93 S.Ct. 2321, 37 L.Ed.2d 298 (1973); Mahan v. Howell, 410 U.S. 315, 93 S.Ct. 979, 35 L.Ed.2d 320 (1973).
In its application of the one person, one vote concept to legislatively-mandated congressional districting, a requirement derived from the constitutional mandate that congressmen be chosen “by the People of the several States,” United States Constitution, article I, section 2, Wesberry v. Sanders, 376 U.S. 1, 84 S.Ct. 526, 11 L.Ed.2d 481 (1964), the Court has disallowed even minor deviations. White v. Weiser, 412 U.S. 783, 93 S.Ct. 2348, 37 L.Ed.2d 335 (1973) (total deviation of 4.13% invalid); Kirkpatrick v. Preisler, 394 U.S. 526, 89 S.Ct. 1225, 22 L.Ed.2d 519 (1969) (total deviation of 5.97% invalid).
In other types of redistricting, some leeway is afforded to state legislatures in devising their reapportionment plans. “ ‘[U]nder 10%’ deviations .. . [have been] considered to be of prima facie constitutional validity ... in the context of legislatively enacted apportionments.” Connor v. Finch, 431 U.S. 407, 418, 97 S.Ct. 1828, 1835, 52 L.Ed.2d 465, 475 (1977). See Gaffney v. Cummings, 412 U.S. 735, 93 S.Ct. 2321, 37 L.Ed.2d 298 (1973) (7.83% total deviation from the population norm held valid); White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973) (9.9% total deviation from population norm upheld).
Legislative deviation from equality is, therefore, permissible if based upon “legitimate considerations incident to the effectuation of a rational state policy. . . . ” Reynolds v. Sims, 377 U.S. at 579, 84 S.Ct. at 1391, 12 L.Ed.2d at 537. Some deviation is permitted for purposes of administrative convenience, adherence to historical or geographic boundaries and recognition of separate political units. See, e. g., Mahan v. Howell, 410 U.S. 315, 93 S.Ct. 979, 35 L.Ed.2d 320 (1973) (16.4% deviation approved); Abate v. Mundt, 403 U.S. 182, 91 S.Ct. 1904, 29 L.Ed.2d 399 (1971) (11.9% deviation upheld). See generally Comment, Fair and Effective Representation: Power to the People, 26 Hastings L.J. 190 (1974). Comment, Political Representation: The Search for Judicial Standards, 43 Brooklyn L.Rev. 431, 435-50 (1976).
*1159In devising a reapportionment plan a court is held to equitable standards of voting equality more stringent than those governing a legislature. Connor v. Finch, 431 U.S. 407, 414, 97 S.Ct. 1828, 1833, 52 L.Ed.2d 465, 473; Marshall v. Edwards, 582 F.2d 927, 930 (5th Cir. 1978), cert. denied, 442 U.S. 909, 99 S.Ct. 2820, 61 L.Ed.2d 274 (1979). Although the mathematical precision required for congressional districting is not a prerequisite, unless there are persuasive justifications, articulated by the court, a judicially-mandated reapportionment plan must ordinarily achieve the goal of population equality “with little more than de minimis variation. Chapman v. Meier, 420 U.S. 1, 26-27 & n.19, 95 S.Ct. 751, 766 & n.19, 42 L.Ed.2d 766, 784 & n.19; Connor v. Finch, 431 U.S. at 414, 97 S.Ct. at 1833, 52 L.Ed.2d at 473; Connor v. Coleman, 440 U.S. 612, 619, n.4, 99 S.Ct. 1523, 1527 n.4, 59 L.Ed.2d 619, 626 n.4 (1979) (Marshall, J., dissenting) (court-ordered plan given less deference than legislative apportionments as to variances from population equality).
By this standard, Plan 1,’ even with the modifications suggested as possible by Mr. Selle, would have resulted in a deviation of 8.2%, far more than de minimis. We conclude that Plan 1, as a court-ordered plan, would have allowed a deviation greater than is permitted by equality of the vote principles apart from any considerations of race, while the total deviation factor under Plan 3 is sufficiently de minimis to satisfy the one person, one vote precepts as applied to court-ordered plans.
III.
The satisfaction of the quantitative one person, one vote requirement does not insure equality in all the aspects of political representation. The quality of representation is, therefore, subject to challenge if the apportionment scheme operates to minimize or cancel out the voting potential of racial or ethnic minorities.
Even before its decision in Baker v. Carr, the Supreme Court had held that the fifteenth amendment condemns racial gerrymandering. Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960). The Court later held that the fourteenth amendment also proscribes any legislative apportionment the purpose of which is invidiously to minimize the voting strength of racial elements of the population. White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973); Whitcomb v. Chavis, 403 U.S. 124, 91 S.Ct. 1858, 29 L.Ed.2d 363 (1971); Burns v. Richardson, 384 U.S. 73, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966); Fortson v. Dorsey, 379 U.S. 433, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965).
We have applied the proscription against minority vote dilution to local government units including parishes and counties, Zimmer v. McKeithen, 467 F.2d 1381 (5th Cir. 1972), reversed and remanded on other grounds, 485 F.2d 1297 (5th Cir. 1973) (en banc), aff’d on other grounds, sub nom. East Carroll Parish School Board v. Marshall, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296 (1976); Kirksey v. Board of Supervisors of Hinds County, 554 F.2d 139 (5th Cir.) (en banc), cert. denied, 434 U.S. 968, 98 S.Ct. 512, 54 L.Ed.2d 454 (1977); and municipalities, Lipscomb v. Wise, 551 F.2d 1043 (5th Cir. 1977), rev’d on other grounds, 437 U.S. 535, 98 S.Ct. 2493, 57 L.Ed.2d 411 (1978); Nevett v. Sides, 571 F.2d 209 (5th Cir. 1978). Cf. Wise v. Lipscomb, 437 U.S. 535, 549, 98 S.Ct. 2493, 2502, 57 L.Ed.2d 411, 422 (1978) (Rehnquist, J., concurring) (four justices express the view that the applicability of minority voting strength dilution doctrine to municipal governments is unsettled). See generally Bonapfél, Minority Challenges to At-Large Elections: The Dilution Problem, 10 Ga.L.Rev. 353 (1976); Note, 87 Harv.L. Rev. 1851 (1974).
However, neither the fifteenth nor the fourteenth amendment commands proportional racial or ethnic representation. City of Mobile v. Bolden, 446 U.S. 55, 63 & 75, 100 S.Ct. 1490, 1498 & 1504, 64 L.Ed.2d 47, 56 & 63 (1980). “The Constitution . .. does not demand that each cognizable element of a constituency elect representatives in proportion to its voting strength. White v. Regester [412 U.S. at 765-66, 93 S.Ct. at *11602339, 37 L.Ed.2d at 324]. . . . ” Nevett v. Sides, 571 F.2d 209, 216 (5th Cir. 1978), cert. denied, 446 U.S. 951, 100 S.Ct. 2916, 64 L.Ed.2d 807 (1980). In the absence of an invidious purpose, a state is constitutionally free to draw political boundaries in any manner it chooses. City of Mobile v. Bol-den, 446 U.S. at 61, 100 S.Ct. at 1497, 64 L.Ed.2d at 54. See Gomillion v. Lightfoot, 364 U.S. at 347, 81 S.Ct. at 130, 5 L.Ed.2d at 116 (state’s power to reapportion is absolute absent an unconstitutional condition).
In drawing district lines, a legislative body is forbidden to discriminate invidiously, but it is not required to be color blind. “[N]either the Fourteenth nor the Fifteenth Amendment mandates any per se rule against using racial factors in districting and apportionment.” United Jewish Organizations v. Carey, 430 U.S. 144, 161, 97 S.Ct. 996, 1007, 51 L.Ed.2d 229, 243 (1977). Therefore, a legislative body may consider religious, ethnic or racial factors so long as it does not act with invidious purpose or violate one person, one vote precepts.
IV.
If a legislative body draws the district lines of a voting plan in a manner that discriminates purposely against black voters, the voting plan must be revised to eliminate that defect. It is unnecessary to the resolution of the issues here presented to evaluate all of the kinds of evidence that might demonstrate the invidious purpose. See City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47 (1980). See also Note, Group Representation and Race-Conscious Apportionment: The Roles of States and the Federal Courts, 91 Harv.L. Rev. 1847, 1848-49 (1978). The parochial bodies in Madison Parish had in the past discriminated against black voters by denying them equal access to the political process. Parochial officials redrew court-ordered voting district lines without court sanction and without the approval required under the Voting Rights Act. The result was to transfer five percent of the parish population, all black, from one district to another. Viewed against the background of past practice in the parish, this transfer of black voters from a ward where they would have major effect to one where their electoral impact would be minor suffices for an inference of discriminatory purpose. See City of Mobile v. Bolden, 446 U.S. 55, 100 S.Ct. 1490, 64 L.Ed.2d 47 (1980). See also Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir. 1973) (en banc); McIntosh County Branch of the NAACP v. City of Darien, 605 F.2d 753 (5th Cir. 1979); Cross v. Baxter, 604 F.2d 875 (5th Cir. 1979).
The parochial legislative bodies have thus violated the court order and the Voting Rights Act and failed to fulfill their apportionment responsibilities. Therefore, the federal district court was obliged to take measures to assure the constitutionality of the election process.
As the district court recognized, it is necessary for the voting plan adopted by the court to avoid racial vote dilution. Yet a court is forbidden to take into account the purely political considerations that might be appropriate for legislative bodies. While racial fairness is required of the plan, whether it be legislative or judicial, a plan that assumes block voting and assures proportional representation on that assumption is not. The Supreme Court “has sternly set its face against the claim, however phrased, that, the Constitution somehow guarantees proportional representation.” City of Mobile v. Bolden, 446 U.S. 55, 78, 100 S.Ct. 1490, 1506, 64 L.Ed.2d 47, 65 (1980).
In Marshall v. Edwards, 582 F.2d at 934-36, Judge Wisdom, speaking for this court, eloquently articulated the reasons for the prohibition against redistricting to assure racially proportionate representation. While some democracies provide for proportional representation of parties and ethnic groups, this division has never been an American tradition. We know, too, that racial and ethnic groups do not always vote in solid phalanxes and that the realities of partisan politics may enable a minority in some circumstances to exact more political concessions by swinging its vote to one of two candidates whose majority-race follow*1161ing is approximately equal than it could by electing a candidate of its own identity. See United States v. Board of Supervisors of Forrest County, 571 F.2d 951, 956 (5th Cir. 1978) (“A court-ordered racial gerrymander which would assure that blacks form a sizeable electoral majority in a single district may not be nearly as effective a guarantee of access as the creation of two or more districts with substantial black voter populations such that all candidates in those districts must be responsive to the needs ... of the black electorate.”).
Even as a remedial measure, court plans should not aim at proportional representation. The decisions of the Supreme Court admonish lower federal courts “to act cautiously in reapportionments and to leave racially proportional representation to legislative bodies, at least in the absence of some impelling reason to take it into account, for example, where the correction of historic racial discrimination and not merely proper representation is involved.” Marshall v. Edwards, 582 F.2d at 936.
In the present case, we do not find a need to correct historic racial discrimination, which was alleviated by the prior court order, sufficient to warrant mandating the kind of proportional representation that is alien to our political tradition. While the plan chosen involves some elements of correction of historical discrimination, we note again, as we did in Marshall v. Edwards, that differences exist between racial proportionate redistricting and remedial racial goals in other fields. Voting choices are different from employment data. It may be logically assumed that, if thirty percent of the group of persons having certain skills are black, then, absent discrimination, thirty percent of those employed to utilize those skills would be black. Such assumptions cannot be accurately made in regard to voter preferences. On the one hand, voter choices are not only based on a candidate’s race but also on subjective evaluation of economic, political and moral factors. On the other, if voters desire to elect a candidate solely because of race, religion, or even his prejudice, they are permitted to do so.
Thus, while race may be considered as a factor, safe seats for the minorities are not required of a reapportionment plan in a dilution ease. United States v. Board of Supervisors of Forrest County, 571 F.2d 951, 955 (5th Cir. 1978); Marshall v. Edwards, 582 F.2d at 936. Therefore, the district court was not required to adopt Plan 1 merely because it would, assuming voting along.racial lines, assure four “safe” seats for black candidates, while Plan 3 would provide only three districts in which the black population majority is sufficiently large to guarantee the election of a black candidate.
V.
In making reapportionment decisions, many data may be relevant. It is patent that equal protection demands equal treatment of electors. The early jurisprudence did not define the statistical base that should be used either to determine when inequity exists or to apportion equally. Total population was used as a measure in many cases. See e. g., Chapman v. Meier, 420 U.S. 1, 95 S.Ct. 751, 42 L.Ed.2d 766 (1975); White v. Regester, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314 (1973); Abate v. Mundt, 403 U.S. 182, 91 S.Ct. 1904, 29 L.Ed.2d 399 (1971). Nevertheless, “the Equal Protection Clause does not require the States to use total population figures ... as the standard by which . . . substantial population equivalency is ... measured.” Burns v. Richardson, 384 U.S. 73, 91, 86 S.Ct. 1286, 1296, 16 L.Ed.2d 376, 390 (1966). The Court in Reynolds v. Sims, referred to voter population and citizen population as well as to total population, making no distinction as to the acceptability of a population equality test based on any of the three measures. Burns v. Richardson, 384 U.S. at 91-92, 86 S.Ct. at 1296, 16 L.Ed.2d at 390. Although the number of registered voters was rejected as a sole determinant, voter registration statistics were considered a valid basis for apportionment where their use produced a distribution not differing substantially from the *1162use of population data. Burns v. Richardson, 384 U.S. at 92-93, 86 S.Ct. at 1297, 16 L.Ed.2d at 390. See Ely v. Klahr, 403 U.S. 108, 114 n.7, 91 S.Ct. 1803, 1807-08 n.7, 29 L.Ed.2d 352, 357 n.7 (1971).
However, the Supreme Court has recognized that, “if it is the weight of a person’s vote that matters, total population-even if stable and accurately taken-may not actually reflect that body of voters whose votes must be counted and weighed for purposes of reapportionment, because ‘census persons’ are not voters.” Gaffney v. Cummings, 412 U.S. 735, 746, 93 S.Ct. 2321, 2328, 37 L.Ed.2d 298, 308 (1973). Thus, we noted in Zimmer v. McKeithen, 467 F.2d at 1302, that “the concept of population ... is not possessed of any talismanic quality” because census figures include aliens, nonresidents, children and other persons ineligible to vote. Zimmer v. McKeithen, 485 F.2d at 1302 n.11.
We have pointed to voting age population as an important factor to be considered in assessing the constitutionality of a reapportionment plan. Marshall v. Edwards, 582 F.2d at 937 n.10. Such data were considered by the Supreme Court in United Jewish Organizations v. Carey, 430 U.S. 144, 163-164, 97 S.Ct. 996, 1009, 51 L.Ed.2d 229, 244 (1977) (constitutionality of actions taken to comply with Voting Rights Act) and Beer v. United States, 425 U.S. 130, 135 n.4, 96 S.Ct. 1357, 1360 n.4, 47 L.Ed.2d 629, 636 n.4 (1976) (application of Section 5 of Voting Rights Act of 1965), as well as by this court in Kirksey v. Board of Supervisors of Hinds County, 554 F.2d 139, 141, 149-150 (5th Cir.) (en banc), cert. denied, 434 U.S. 968, 98 S.Ct. 512, 54 L.Ed.2d 454 (1977); and in Moore v. Leflore County Board of Election Commissioners, 502 F.2d 621, 626 (5th Cir. 1974). Cf. City of Rome v. United States, 446 U.S. 156, 186 n.22, 100 S.Ct. 1548, 1567 n.22, 64 L.Ed.2d 119, 145 n.22 (1980) (current voting-age population data more probative in a Voting Rights Act case than are voter registration data).
In Marshall v. Edwards, we rejected a court-ordered plan that had been formulated based on voter registration data alone by the same district court that considered this case. 582 F.2d at 927. Our decision in Marshall v. Edwards apparently prompted the rejection by the district court of the proffer of evidence made here. However, neither the Supreme Court nor this court has ever held that voter registration is irrelevant: it is simply not the sole criterion. More important the proffer here included evidence tending to show a disparity in voting age population. As we have noted, that is an important factor, perhaps the single most significant one in determining vote dilution. Thus, the district court erroneously refused the proffered evidence.
VI.
“Legislators are elected by voters, not farms or cities or economic interests.” Reynolds v. Sims, 377 U.S. at 562, 84 S.Ct. at 1382, 12 L.Ed.2d at 527. Yet, in creating districts, the legislative body may “legitimately desire to maintain the integrity of various political subdivisions, insofar as possible, and provide for compact districts of contiguous territory. ...” Reynolds v. Sims, 377 U.S. at 578, 84 S.Ct. at 1390, 12 L.Ed.2d at 536. It may consider natural or historical boundary lines. 377 U.S. at 578-79, 84 S.Ct. at 1390, 12 L.Ed.2d at 536. The overriding objectives must be substantial equality of population among the various districts, without the dilution of the strength of minority voting groups. Geographical and administrative considerations are not reprobated so long as they are not used to justify disparity from equal and not invidiously biased representation. Moore v. Leflore County Board of Election Commissioners, 502 F.2d 621, 625 (5th Cir. 1974).
These factors are as appropriate for consideration in a court-ordered plan. However, the record in the present case does not contain either specific findings of fact or sufficient evidence to enable us to review the district court’s determination. While judicial notice can be taken of evidence not in the record, Fed.R.Evid. 201, the district court opinion contains no indication that this was done.
We cannot, therefore, determine whether, as urged by the plaintiffs, the creation of *1163three predominantly rural districts, as suggested in Plan 3, properly meets historical, administrative, geographic and legitimate planning objectives, or whether the court’s emphasis on preserving rural interests in drawing district lines is an abuse of discretion either because it fails to meet these criteria or because, even while doing so, it defeats the primary dual tests of voter equality and racial fairness.
VII.
District courts are not denied discretion in adopting voting plans. Indeed, because of the almost infinite variety of patterns such plans may take, appellate courts must and do impart considerable discretion to trial judges. In this opinion, we have tried to set forth the factors that must guide the trial judge. We are aware that many of the strictures are negative and that “thou-shalt-nots” are much less helpful than affirmative prescriptions. However, we cannot reduce to a formula the factors that require the consideration of the district court. The district judge must steer a middle course, following the guidance afforded by Marshall v. Edwards. He must be mindful of the impact of a proposed plan on different racial groups. He must analyze the plan and determine that it does not dilute minority voting strength but he must avoid a strict proportionality brought about by the manipulation of district lines. He should fix boundaries that are compact, contiguous and that preserve natural, political and traditional representation. Beyond that courts should not go: “We are not legislatures.” Marshall v. Edwards, 582 F.2d at 937.
VIII.
We conclude that the district court should have accepted the proffer of evidence to show both voter registration and voting age population. Although no one offered such evidence, it would have been desirable for the court to have precise data on voting age population as an important basis for its plan. In any event, Plan 1 was not acceptable because of its proportional representation premise. Although Plan 3 embodies an acceptable deviation factor, we might consider setting aside the decree adopting Plan 3 and remanding for consideration of the excluded evidence and the adoption of a plan taking this opinion into account.
Events, however, have not abided our decision. Elections have been held. No new elections are scheduled in the immediate future. Compare Watson v. Commissioners Court of Harrison County, 616 F.2d 105 (5th Cir. 1980). More important, an official census has been taken and its results will shortly be released. Not only is this likely to provide more complete data than a census by a special master, it will be available without additional expense. See Ely v. Klahr, 403 U.S. 108, 114, 91 S.Ct. 1803, 1807, 29 L.Ed.2d 352, 356 (1971) (district court “did not err in giving the legislature a reasonable time to act based on the 1970 census figures which the court thought would be available in the summer of 1971”). The census data may require legislative reapportionment under Louisiana law. Any such reapportionment would, of course, be subject to the requirements of Section 5 of the Voting Rights Act, 42 U.S.C. § 1973c.
Accordingly, we let stand the district court’s judgment adopting reapportionment Plan 3. However, if within six months from the date the 1980 census data for Madison Parish is officially published, the Police Jury and School Board have not reapportioned themselves, the plaintiffs may apply for further relief. Should the Police Jury and School Board adopt a plan of reapportionment, the plaintiffs may, of course, challenge that plan. Should census data not be available at least six months prior to the next scheduled election, the plaintiffs may apply for a stay or such other interim relief as may be appropriate. The criteria set forth in this opinion shall guide the court in the event its further intercession is sought.
For these reasons the judgment is AFFIRMED.
. Plan 1
Median District Population 1806
Deviation District Population from Median Population Percentage Percentage
1 1959 + 8.4% 60% 40%
2 1826 + 1.1% 67% 33%
3 1752 -2.9% 7%
4 1746 -3.3% 1%
5 1920 + 3.3%
6 1789 100%
7 1805 2%
8 1910 100%
. Plan 3
Median District Population 1806
Deviation from Median White Black Population Population Percentage Percentage
1 1772 -1.9 47% 53%
2 1818 + .66 75% 25%
3 1803 62%
4 1808 100%
5 1789 100%
1846 + 2.21% 93%
7 1799 - .38% 65% 35%
8 1842 + 1.9 % 100%
. The explanation given was that most of the family units in the housing project consist of a single adult with several children. Therefore, a large percentage of the black population is not old enough to register.
. Mr. Selle testified that Plan 1 could be modified to reduce the deviation factor from 11.7% to 8.2%.