This action involves a challenge to the apportionment1 of the electoral districts for the Board of Commissioners and Board of Education of Mecklenburg County, North Carolina. Plaintiffs-Appellees claim that the voting districts at issue violate the constitutional principle of one person, one vote because the voting-age populations of the districts are not substantially equal. In other words, Plaintiffs claim that the districting plan violates the Equal Protection Clause because each voter in a district with fewer eligible voters has a greater voice in electing a representative than does a voter in a district with more eligible voters.

On cross-motions for summary judgment, the district court ruled that the districting plan is unconstitutional because the deviation among the voting-age populations of the districts exceeds the maximum allowable level. The court also determined that Defendants-Appellants offered no legitimate justification for the large variances in voting-age population among the districts. Accordingly, the court granted summary judgment for Plaintiffs and enjoined Defendants from conducting elections under the challenged legislation.

Defendants appeal the district court's order. They claim that the district court erred in using voting-age population instead of total population as the basis for comparing the voting districts. For the reasons that follow, we vacate the district court's ruling and remand for further proceedings.I.

The material facts of this case are essentially undisputed. Prior to 1990, the Mecklenburg County Board of Commissioners had seven members: four were elected from single-member districts, and three were elected at large. In addition, at that time the Mecklenburg County Board of Education had nine members, all elected at large.

The federal decennial census of 1990 indicated that the boundaries of the four single-member districts of the Board of Commissioners needed to be redrawn because of shifts in population. Also, about that time the public began to express interest in enlarging the size of the Board of Commissioners from seven members to nine. In response, the Board appointed a committee, known as the Blue Ribbon County Governance Committee, to provide recommendations about the size of the Board and the boundaries for the voting districts. The Blue Ribbon Committee recommended increasing the Board of Commissioners from seven members to nine, six elected from single-member districts and three elected at large. The Committee also proposed a districting map depicting the boundaries of the six single-member districts. The Board approved the Committee's recommendations and placed the proposal on the 1992 general election ballot. The voters approved the changes effective for the 1994 election cycle.

After the changes to the Board of Commissioners were approved, but before they were implemented, public interest also arose about changing the method of electing members of the Board of Education, from the existing all-at-large system to a district system similar to that used for the Board of Commissioners. In the 1993 session of the North Carolina General Assembly, Senate Bill 613 was introduced, which provided that six members of the Board of Education would be elected from single-member districts and the remaining three members would be elected at-large. In addition, the bill revised the newly created districts for the Board of Commissioners and provided that those same districts would also be used for the Board of Education. Senate Bill 613 was enacted by the General Assembly on June 16, 1993 and approved by the voters of Mecklenburg County on November 2, 1993.

Plaintiffs-Appellees filed suit on November 8, 1993 challenging the election districts established by Senate Bill 613 for the Board of Commissioners and the Board of Education. All Plaintiffs are residents and registered voters of Mecklenburg County, North Carolina. They allege that the new voting districts violate the one person, one vote principle because the populations of the districts are not sufficiently equal. According to Plaintiffs, the initial reapportionment plan for the Board of Commissioners that was drafted by the Blue Ribbon Committee yielded a maximum deviation of only 1.55% in terms of total population.2 In comparison, the reapportionment plan of Senate Bill 613 increased the maximum deviation to 8.33% in terms of total population. J.A. at 74a. Plaintiffs claim that this increase demonstrates that the new districting plan was not the result of a good-faith effort to achieve districts as nearly of equal population as is practicable. Moreover, Plaintiffs' principal argument in support of their claim of vote dilution is that the maximum deviation of 16.17% in terms of voting-age population among the districts under the new plan3 is unconstitutionally large.

Both parties moved for summary judgment. On March 30, 1995, the district court issued an order granting Plaintiffs' motion for summary judgment, denying Defendants' motion, and enjoining Defendants from conducting elections under the challenged plan. The district court determined that the districting plan violated the Equal Protection Clause because the deviation of 16.17% in terms of voting-age population exceeds the maximum allowable deviation of 10%.

Defendants appealed. They claim that the district court erred in using voting-age population, rather than total population, as the proper criterion for determining compliance with the one person, one vote principle in this action. Defendants contend that they are entitled to summary judgment because the stipulated maximum deviation of 8.33% in terms of total population is de minimis as a matter of law.

Plaintiffs respond that the district court was correct in using voting-age population because that figure more accurately represents actual voting strength in the districts. In addition, Plaintiffs claim on appeal that even if total population is used, they can nevertheless maintain a viable case of vote dilution under the Equal Protection Clause even when the maximum population deviation is less than 10%.

The issues before the court present questions of law and are therefore subject to de novo review. Duvall v. Bristol-Myers-Squibb Co., 65 F.3d 392, 395 (4th Cir.1995), cert. granted, --- U.S. ----, 116 S. Ct. 2575, 135 L. Ed. 2d 1090 (1996). We address these issues below after a brief overview of the Supreme Court's one person, one vote cases.

The equal protection guarantee of "one person, one vote"4 requires that representatives to an elected body are elected from voting districts of substantially equal population. Reynolds v. Sims, 377 U.S. 533, 577, 84 S. Ct. 1362, 1389-90, 12 L. Ed. 2d 506 (1964). This principle ensures that every voter, no matter what district he or she lives in, will have an equal say in electing a representative. It also ensures that every person receives equal representation by his or her elected officials. The United States Supreme Court has applied the one person, one vote principle to elections for congressional representatives, Wesberry v. Sanders, 376 U.S. 1, 84 S. Ct. 526, 11 L. Ed. 2d 481 (1964);5 state legislative representatives, Reynolds v. Sims, supra; and local governmental representatives, Avery v. Midland County, 390 U.S. 474, 88 S. Ct. 1114, 20 L. Ed. 2d 45 (1968).

In the landmark apportionment case of Reynolds v. Sims, the Supreme Court established that the one person, one vote principle is inherent in the Equal Protection Clause of the Fourteenth Amendment. The Reynolds Court recognized that "the fundamental principle of representative government in this country is one of equal representation for equal numbers of people, without regard to race, sex, economic status, or place of residence within a State." 377 U.S. at 560-61, 84 S.Ct. at 1381. Accordingly, the Court held that "the Equal Protection Clause requires that a State make an honest and good faith effort to construct districts, in both houses of its legislature, as nearly of equal population as is practicable." Id. at 577, 84 S.Ct. at 1390.

The Court in Reynolds realized, however, "that it is a practical impossibility to arrange legislative districts so that each one has an identical number of residents, or citizens, or voters. Mathematical exactness or precision is hardly a workable constitutional requirement." Id. Since Reynolds, the Court has refined the parameters for determining when population differences among legislative districts unconstitutionally dilute the votes of the members of the larger districts. In Gaffney v. Cummings, 412 U.S. 735, 93 S. Ct. 2321, 37 L. Ed. 2d 298 (1973), the Court held that "minor deviations from mathematical equality among state legislative districts are insufficient to make out a prima facie case of invidious discrimination under the Fourteenth Amendment so as to require justification by the State." Id. at 745, 93 S.Ct. at 2327. Similarly, in White v. Regester, 412 U.S. 755, 93 S. Ct. 2332, 37 L. Ed. 2d 314 (1973), the Court stated, "we do not consider relatively minor population deviations among state legislative districts to substantially dilute the weight of individual votes in the larger districts so as to deprive individuals in these districts of fair and effective representation." Id. at 764, 93 S.Ct. at 2338.6

The Court has come to recognize what is generally regarded as a benchmark for determining whether a particular apportionment plan violates the one person, one vote principle. In Brown v. Thomson, 462 U.S. 835, 103 S. Ct. 2690, 77 L. Ed. 2d 214 (1983), the Court stated, "Our decisions have established, as a general matter, that an apportionment plan with a maximum population deviation under 10% falls within this category of minor deviations. A plan with larger disparities in population, however, creates a prima facie case of discrimination and therefore must be justified by the State." Id. at 842-43, 103 S.Ct. at 2696 (citations omitted); cf. Connor v. Finch, 431 U.S. 407, 418, 97 S. Ct. 1828, 1835, 52 L. Ed. 2d 465 (1977) ("The maximum population deviations of 16.5% in the Senate districts and 19.3% in the House districts can hardly be characterized as de minimis; they substantially exceed the 'under-10%' deviations the Court has previously considered to be of prima facie constitutional validity only in the context of legislatively enacted apportionments."). Thus, the Court's one person, one vote cases have generally established three levels for gauging whether population deviations between voting districts violate the Equal Protection Clause.7 If the maximum deviation is less than 10%, the population differential will be considered de minimis and will not, by itself, support a claim of vote dilution. If the maximum deviation is greater than 10%, it is prima facie evidence of a one person, one vote violation, and the state must justify the population disparity by showing a rational and legitimate state policy for the districting plan.8 Brown, 462 U.S. at 842-43, 103 S.Ct. at 2695-96. Finally, there is a level of population disparity beyond which a state can offer no possible justification. Although it is not clear precisely what that upper level is, the Court has stated in dictum that a maximum deviation of 16.4% "may well approach tolerable limits." Mahan v. Howell, 410 U.S. 315, 329, 93 S. Ct. 979, 987, 35 L. Ed. 2d 320 (1973).

As an initial matter, Plaintiffs in this action argue that they can establish a viable case of vote dilution even if the maximum deviation is less than 10%. They assert that the Supreme Court has never firmly established the 10% level as an absolute bright-line test of the constitutionality of a state legislative apportionment plan.

First, Plaintiffs contend that Brown v. Thomson, supra, which is often cited as acknowledging the 10% rule, was merely a plurality opinion. This contention is simply wrong. Although Justice O'Connor, joined by Justice Stevens, filed a separate concurring opinion in Brown, they both joined completely in Justice Powell's opinion for the Court. See 462 U.S. at 836, 103 S.Ct. at 2692-93 ("Powell, J., delivered the opinion of the Court, in which Burger, C.J., and Rehnquist, Stevens, and O'Connor, JJ., joined.").

Justice O'Connor does state, in her concurring opinion in Brown, that "ensuring equal representation is not simply a matter of numbers. There must be flexibility in assessing the size of the deviation against the importance, consistency, and neutrality of the state policies alleged to require the population disparities." Id. at 848, 103 S.Ct. at 2699 (O'Connor, J., concurring). However, the context of this statement indicates that she is not denouncing the 10% threshold required to establish a prima facie case of vote dilution. Rather, it is apparent that Justice O'Connor's statement about the need for flexibility refers only to cases in which the plaintiff has already established a prima facie case, because only then is a state required to justify the disparity. In fact, later in her opinion, she acknowledges that "[i]n the past, this Court has recognized that a state legislative apportionment scheme with a maximum population deviation exceeding 10% creates a prima facie case of discrimination." Id. at 850, 103 S.Ct. at 2700.

Furthermore, even the dissenting justices in Brown appear to have acknowledged the 10% de minimis benchmark:

Our cases since Reynolds have clarified the structure of constitutional inquiry into state legislative apportionments, setting up what amounts to a four-step test. First, a plaintiff must show that the deviations at issue are sufficiently large to make out a prima facie case of discrimination. We have come to establish a rough threshold of 10% maximum deviation from equality (adding together the deviations from average district size of the most underrepresented and most overrepresented districts); below that level, deviations will ordinarily be considered de minimis.

462 U.S. at 852, 103 S. Ct. at 2701 (Brennan, J., dissenting).

Next, Plaintiffs assert that the Supreme Court never intended for the 10% maximum population deviation level to serve as a "safe harbor" below which a state's districting scheme is immune from attack. According to Plaintiffs, Justice Powell's opinion in Brown was merely an observation that the Court's prior cases, when decided on an individual basis, have divided themselves roughly at a de minimis level of 10%. Plaintiffs assert that the Brown Court's use of the phrase "as a general matter" indicates that the 10% guideline is not automatically dispositive. According to Plaintiffs, every alleged violation of the one person, one vote principle must be analyzed on an individual, case-by-case basis.

Plaintiffs' interpretation does enjoy some support among the Court's prior one person, one vote decisions. For example, in Roman v. Sincock, 377 U.S. 695, 84 S. Ct. 1449, 12 L. Ed. 2d 620 (1964), decided the same day as Reynolds v. Sims, the Court rejected the district court's attempt to establish a fixed mathematical formula for judging the constitutionality of population variances among voting districts. The Roman Court stated,

[T]he problem does not lend itself to any such uniform formula, and it is neither practicable nor desirable to establish rigid mathematical standards for evaluating the constitutional validity of a state legislative apportionment scheme under the Equal Protection Clause. Rather, the proper judicial approach is to ascertain whether, under the particular circumstances existing in the individual State whose legislative apportionment is at issue, there has been a faithful adherence to a plan of population-based representation, with such minor deviations only as may occur in recognizing certain factors that are free from any taint of arbitrariness or discrimination.

377 U.S. at 710, 84 S. Ct. at 1458.

In addition, in Kirkpatrick v. Preisler, 394 U.S. 526, 89 S. Ct. 1225, 22 L. Ed. 2d 519 (1969), the Court rejected the argument that there is a precise mathematical benchmark below which population variances are automatically considered de minimis. The Court had previously held that "Art. I, § 2 of the Constitution requires that 'as nearly as is practicable one man's vote in a congressional election is to be worth as much as another's.' " Id. at 527-28, 89 S.Ct. at 1227 (quoting Wesberry v. Sanders, 376 U.S. 1, 7-8, 84 S. Ct. 526, 530, 11 L. Ed. 2d 481 (1964)). According to the Kirkpatrick Court,

The whole thrust of the "as nearly as practicable" approach is inconsistent with adoption of fixed numerical standards which excuse population variances without regard to the circumstances of each particular case. The extent to which equality may practicably be achieved may differ from State to State and from district to district. Since "equal representation for equal numbers of people [is] the fundamental goal for the House of Representatives," the "as nearly as practicable" standard requires that the State make a good-faith effort to achieve precise mathematical equality. Unless population variances among congressional districts are shown to have resulted despite such effort, the State must justify each variance, no matter how small.

394 U.S. at 530-31, 89 S. Ct. at 1228-29 (citations omitted). Also, the Court in Kirkpatrick explained some valid practical reasons for rejecting a fixed de minimis level: "We can see no nonarbitrary way to pick a cutoff point at which population variances suddenly become de minimis. Moreover, to consider a certain range of variances de minimis would encourage legislators to strive for that range rather than for equality as nearly as practicable." Id. at 531, 89 S.Ct. at 1229.

Furthermore, in Swann v. Adams, 385 U.S. 440, 87 S. Ct. 569, 17 L. Ed. 2d 501 (1967), the Court stated, "the fact that a 10% or 15% variation from the norm is approved in one State has little bearing on the validity of a similar variation in another State. 'What is marginally permissible in one State may be unsatisfactory in another, depending on the particular circumstances of the case.' " Id. at 445, 87 S.Ct. at 572 (quoting Reynolds v. Sims, 377 U.S. at 578, 84 S.Ct. at 1390); see also Mahan v. Howell, 410 U.S. 315, 329, 93 S. Ct. 979, 987, 35 L. Ed. 2d 320 (1973) ("Neither courts nor legislatures are furnished any specialized calipers that enable them to extract from the general language of the Equal Protection Clause of the Fourteenth Amendment the mathematical formula that establishes what range of percentage deviations is permissible, and what is not.").

The above quotes from Roman, Kirkpatrick, Swann, and Mahan certainly appear to indicate that the Court would be hesitant to establish a bright-line test in one person, one vote cases; however, those cases are not dispositive here. First, Kirkpatrick involved a challenge to a congressional apportionment plan. As indicated in footnote 5, supra, the equal population requirements for congressional districts, which are imposed by Art. I, § 2 of the Constitution, are more stringent than those for state or local legislative districts, which are governed by the Equal Protection Clause of the Fourteenth Amendment.

Second, the Court's apportionment decisions subsequent to Roman, Swann, and Mahan indicate the Court's willingness to recognize a de minimis level below which population variances are deemed acceptable. For example, in White v. Regester, the Court expressly stated,

[W]e did not hold in Swann v. Adams, 385 U.S. 440 [87 S. Ct. 569, 17 L. Ed. 2d 501] (1967), or Kilgarlin v. Hill, 386 U.S. 120 [87 S. Ct. 820, 17 L. Ed. 2d 771] (1967), or later in Mahan v. Howell, [410 U.S. 315, 93 S. Ct. 979, 35 L. Ed. 2d 320 (1973) ], that any deviations from absolute equality, however small, must be justified to the satisfaction of the judiciary to avoid invalidation under the Equal Protection Clause. For the reasons set out in Gaffney v. Cummings, supra, we do not consider relatively minor population deviations among state legislative districts to substantially dilute the weight of individual votes in the larger districts so as to deprive individuals in these districts of fair and effective representation.

412 U.S. 755, 763-64, 93 S. Ct. 2332, 2338, 37 L. Ed. 2d 314 (1973).

The 10% de minimis threshold recognized in Brown does not completely insulate a state's districting plan from attack of any type. Instead, that level serves as the determining point for allocating the burden of proof in a one person, one vote case. A maximum deviation of greater than 10% automatically establishes a prima facie violation of the one person, one vote principle. If the plaintiff establishes this level of disparity in population among the districts, the burden of proof shifts to the state to justify the deviations by showing a rational and legitimate state policy for the districts.

On the other hand, if the maximum deviation is less than 10%, the population disparity is considered de minimis and the plaintiff cannot rely on it alone to prove invidious discrimination or arbitrariness. To survive summary judgment, the plaintiff would have to produce further evidence to show that the apportionment process had a "taint of arbitrariness or discrimination." Roman v. Sincock, 377 U.S. at 710, 84 S. Ct. at 1458. In other words, for deviations below 10%, the state is entitled to a presumption that the apportionment plan was the result of an "honest and good faith effort to construct districts ... as nearly of equal population as is practicable." Reynolds v. Sims, 377 U.S. at 577, 84 S. Ct. at 1390. However, this is a rebuttable presumption.

As the Court recognized in Gaffney v. Cummings, "State legislative districts may be equal or substantially equal in population and still be vulnerable under the Fourteenth Amendment." 412 U.S. at 751, 93 S.Ct. at 2330. Indeed, even after the Court in Gaffney found that the deviations at issue there failed to make out a prima facie case of discrimination, the Court addressed the plaintiff's claim that the apportionment plan was invidiously discriminatory because it was based on a "political fairness principle." Presumably, an apportionment plan that satisfies the 10% de minimis threshold could nevertheless be challenged under another theory, such as a violation of the Voting Rights Act or as an unconstitutional racial gerrymander under Shaw v. Reno, 509 U.S. 630, 113 S. Ct. 2816, 125 L. Ed. 2d 511 (1993), and Miller v. Johnson, 515 U.S. 900, 115 S. Ct. 2475, 132 L. Ed. 2d 762 (1995).

Plaintiffs also contend that the existence of other viable districting plans with smaller population discrepancies--i.e., plans that are "more constitutionally perfect"--indicates that the population variances were not unavoidable and, therefore, that the state's apportionment plan was not the result of a good-faith effort to achieve absolute equality. The Supreme Court has expressly rejected the argument that the possibility of drafting a "better" plan alone is sufficient to establish a violation of the one person, one vote principle. In Gaffney v. Cummings, the Court specifically stated:

We think that appellees' showing of numerical deviations from population equality among the Senate and House districts in this case failed to make out a prima facie violation of the Equal Protection Clause of the Fourteenth Amendment, whether those deviations are considered alone or in combination with the additional fact that another plan could be conceived with lower deviations among the State's legislative districts.

412 U.S. at 740-41, 93 S. Ct. at 2325 (emphasis added). Furthermore, the Gaffney Court stated that judicial involvement in the inherently legislative process of apportionment "must end at some point, but that point constantly recedes if those who litigate need only produce a plan that is marginally 'better' when measured against a rigid and unyielding population-equality standard. The point is, that such involvements should never begin." Id. at 750-51, 93 S.Ct. at 2330.

In Swann v. Adams, 385 U.S. 440, 87 S. Ct. 569, 17 L. Ed. 2d 501 (1967), the Court did discuss other proposed districting plans, or simple amendments to the challenged plan, that could have produced smaller population variances among districts. See id. at 445-46, 87 S.Ct. at 572-73. However, the Swann Court did not hold that the mere existence of a plan of better population equality is sufficient to establish a one person, one vote violation. In Swann, the state argued, as one of its justifications for the challenged apportionment plan, that the plan came as close as practicable to complete population equality. The Court rejected this practicality argument because other proposed plans demonstrated that "the State could have come much closer to providing districts of equal population than it did." Id. at 445, 87 S.Ct. at 573. The Court's discussion in Swann of alternative apportionment plans arose only in the context of the state's justification for population deviations in excess of the de minimis level, after the plaintiff had established a prima facie case of vote dilution. In Gaffney, which was decided subsequent to Swann, the Court expressly rejected the argument that a plaintiff can use the possibility of a more equipopulous apportionment plan alone to establish a one person, one vote violation. See Gaffney, 412 U.S. at 735, 93 S.Ct. at 2322.

Plaintiffs next argue that the State of North Carolina's ability to create twelve congressional districts that vary by exactly one person is evidence that the state did not act in good faith in creating the election districts at issue here, which vary by 7,103 people. This argument ignores the Supreme Court's holdings that state and local election districts are not examined under the same criteria as are congressional districts. See Mahan v. Howell, 410 U.S. 315, 321, 93 S. Ct. 979, 983, 35 L. Ed. 2d 320 (1973) ("[M]ore flexibility [is] constitutionally permissible with respect to state legislative apportionment than in congressional redistricting."); Abate v. Mundt, 403 U.S. 182, 185, 91 S. Ct. 1904, 1907, 29 L. Ed. 2d 399 (1971) (suggesting that "slightly greater percentage deviations may be tolerated for local government apportionment schemes").

There are certainly some credible arguments for rejecting the idea that the Supreme Court has established an absolute rule that any apportionment plan with a maximum population deviation below 10% is automatically presumed to be valid. However, Plaintiffs do not cite any decisions that explicitly adopt their interpretation of the apportionment cases. Therefore, this court agrees with the district court that a districting plan with a maximum deviation of less than 10% contains a de minimis level of population variation.

Whether Plaintiffs can produce any credible evidence to establish that the apportionment plan at issue here was the product of bad faith, arbitrariness, or invidious discrimination should be addressed on remand.

The principal issue in this appeal is Defendants' argument that the district court erred in using voting-age population, instead of total population, as the determinative statistic for measuring compliance with the one person, one vote principle in this case. Defendants contend that total population is unquestionably an acceptable means of apportioning legislative districts and measuring alleged vote dilution. As Defendants observe, the Supreme Court has consistently used total population figures, based on federal census data, to analyze one person, one vote cases.

Plaintiffs respond by asserting that the district court properly determined that the apportionment plan was invalid because it produced an unacceptably high deviation in terms of voting-age population among the districts. The district court determined that voting-age population is the more appropriate apportionment base because it provides a better indication of actual voting strength than does total population. The district court's ruling is based primarily on Judge Kozinski's separate opinion in Garza v. County of Los Angeles, 918 F.2d 763 (9th Cir.1990), cert. denied, 498 U.S. 1028, 111 S. Ct. 681, 112 L. Ed. 2d 673 (1991), in which Kozinski analyzed the potential conflict between the principles of electoral equality and representational equality. This theory, as well as the respective arguments of both parties, is explored below.

In Reynolds v. Sims, the Supreme Court stated that "[p]opulation is, of necessity, the starting point for consideration and the controlling criterion for judgment in legislative apportionment controversies." 377 U.S. at 567, 84 S.Ct. at 1384. Although the Reynolds Court's analysis was based on total population, the Court "carefully left open the question what population was being referred to." Burns v. Richardson, 384 U.S. 73, 91, 86 S. Ct. 1286, 1296, 16 L. Ed. 2d 376 (1966). In fact, at one point in Reynolds, the Court recognized the difficulty of apportioning electoral districts "so that each one has an identical number of residents, or citizens, or voters." 377 U.S. at 577, 84 S. Ct. at 1390 (emphasis added).

The Supreme Court has never directly addressed a situation in which an apportionment plan that is acceptable in terms of total population equality nevertheless creates an unacceptable disparity among the districts in terms of actual voter strength. Apparently, the Ninth Circuit Court of Appeals, in Garza v. County of Los Angeles, supra, is the only federal circuit court that has directly faced this question.

In Garza, the County of Los Angeles appealed a district court order that imposed a remedial redistricting plan based on total population, rather than voting-age citizen population. The County argued that using total population unconstitutionally weights the votes of electors who live in districts with high concentrations of alien Hispanics, because the non-citizen residents are counted in total population, but are ineligible to vote. The Garza court upheld the district court's use of total population as the apportionment base. The majority of the court noted, inter alia, that "[t]he purpose of redistricting is not only to protect the voting power of citizens; a coequal goal is to ensure 'equal representation for equal numbers of people.' " 918 F.2d at 775 (quoting Kirkpatrick v. Preisler, 394 U.S. 526, 531, 89 S. Ct. 1225, 1229, 22 L. Ed. 2d 519 (1969)). The court determined that basing voting districts on voter population "would dilute the access of voting age citizens in [the Hispanic] district to their representative, and would similarly abridge the right of aliens and minors to petition that representative." Id.

Judge Kozinski filed a separate opinion in Garza, in which he analyzed the two proposed apportionment bases at issue--total population and elector population--in terms of the different principles of representational government advanced by each method.

According to Kozinski, apportionment by total population furthers the principle of representational equality. This principle ensures that all constituents, whether or not they are eligible to vote, have roughly equal access to their elected representatives to voice their opinions or otherwise to advance their interests. In addition, "assuming that elected officials are able to obtain benefits for their districts in proportion to their share of the total membership of the governing body, it assures that constituents are not afforded unequal government services depending on the size of the population in their districts." Id. at 781 (Kozinski, J., concurring in part and dissenting in part).

On the other hand, apportionment by voter-eligible population serves the principle of electoral equality. This principle

assures that, regardless of the size of the whole body of constituents, political power, as defined by the number of those eligible to vote, is equalized as between districts holding the same number of representatives. It also assures that those eligible to vote do not suffer dilution of that important right by having their vote given less weight than that of electors in another location.

Id. at 781-82.

Kozinski suggested that apportionment by either total population or elector population will normally satisfy both concerns of representational equality and electoral equality. However, under the particular facts presented in Garza, because of the demographic abnormalities in Los Angeles County, he stated that using total population as the apportionment base would not assure equality of voting strength among the districts. Id. at 781. Kozinski analyzed the Supreme Court's prior one person, one vote cases and determined that the overriding concern of those cases is electoral equality. Id. at 782-84. Therefore, he concluded that when the principles of representational equality and electoral equality cannot coexist, electoral equality must prevail. In other words, he determined that in cases where apportionment by total population does not assure equal voter strength, some other criterion, such as voter-eligible population, must be used as the apportionment base. Id. at 784.

This court does not agree with Judge Kozinski, and the district court below here, that the Supreme Court's prior one person, one vote cases suggest that the principle of electoral equality is superior to the principle of representational equality. The Court seems to have assumed that these two principles always go hand in hand. For example, in Reynolds v. Sims, the Court stated, "the overriding objective must be substantial equality of population among the various districts, so that the vote of any citizen is approximately equal in weight to that of any other citizen in the State." 377 U.S. at 579, 84 S.Ct. at 1390 (emphasis added). At another point in Reynolds, the Court refers to "[t]he right of a citizen to equal representation and to have his vote weighted equally with those of all other citizens in the election of [representatives]." Id. at 576, 84 S.Ct. at 1389 (emphasis added). Also, in Board of Estimates v. Morris, 489 U.S. 688, 109 S. Ct. 1433, 103 L. Ed. 2d 717 (1989), the Court stated, "It may be that in terms of assuring fair and effective representation, the equal protection approach reflected in the Reynolds v. Sims line of cases is itself imperfect, but it does assure that legislators will be elected by, and represent citizens in, districts of substantially equal size." Id. at 699, 109 S.Ct. at 1441 (emphasis added).

The Supreme Court's prior opinions offer no clue as to which principle--electoral equality or representational equality--is more important in a democratic society. Although Judge Kozinski cites to numerous passages from the Court's apportionment cases that refer to the right to equally weighted voting, one can find equally compelling support for the principle of representational equality. For example, as noted previously, the Reynolds Court stated that "the fundamental principle of representative government in this country is one of equal representation for equal numbers of people, without regard to race, sex, economic status, or place of residence within a State." 377 U.S. at 560-61, 84 S.Ct. at 1381 (emphasis added). Also in Reynolds, the Court stated, "The Equal Protection Clause demands no less than substantially equal state legislative representation for all citizens, of all places as well as of all races." Id. at 568, 84 S.Ct. at 1385.

Kozinski suggests that the "subservience of the representational principle to the principle of electoral equality" is demonstrated by the Supreme Court's opinion in Gaffney v. Cummings, 412 U.S. 735, 93 S. Ct. 2321, 37 L. Ed. 2d 298 (1973). Garza, 918 F.2d at 783. In Gaffney, the Court recognized that total population is not a perfect proxy for voting strength:

[I]t must be recognized that total population, even if absolutely accurate as to each district when counted, is nevertheless not a talismanic measure of the weight of a person's vote under a later adopted reapportionment plan.... District populations are constantly changing, often at different rates in either direction, up or down. Substantial differentials in population growth rates are striking and well-known phenomena. So, too, 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 the purposes of reapportionment, because "census persons" are not voters. The proportion of the census population too young to vote or disqualified by alienage or nonresidence varies substantially among the States and among localities within the States.

412 U.S. at 746-47, 93 S. Ct. at 2328 (footnotes omitted). Kozinski suggests that this quotation from Gaffney implies that equality of population is not the goal of the one person, one vote cases, but rather a means of achieving the end of electoral equality. Accordingly, argues Kozinski, courts should not necessarily rely on total population figures in one person, one vote cases to assure compliance with electoral equality. Garza, 918 F.2d at 783.

The above-quoted language from Gaffney, however, cuts two ways. Although the Court in Gaffney recognized the inherent inaccuracy of using total population to approximate voting strength, the Court apparently attributed no constitutional significance to the discrepancy. The Gaffney Court used total population figures to analyze the legislative apportionment plan at issue there even after observing that the congressional districts from the same state contained significant variations in terms of age-eligible voters.9 The Court in Gaffney merely recognized that the imprecision of total population as an indicator of actual voter strength requires that courts use some flexibility in performing a one person, one vote analysis. The Court suggested that a reasonable margin of error should be allowed so that courts would not assume that voter strength was unconstitutionally disparate simply because total population was not exactly equal.

Judge Kozinski also relied heavily on the Supreme Court's opinion in Burns v. Richardson, 384 U.S. 73, 86 S. Ct. 1286, 16 L. Ed. 2d 376 (1966), to support his conclusion that electoral equality, not representational equality, is the essence of one person, one vote. Kozinski suggests that Burns is very important to this analysis because it is the only Supreme Court case applying the one person, one vote principle in a situation where there was a divergence between representational equality and electoral equality. Garza, 918 F.2d at 783-84.

In Burns, the Court approved Hawaii's use of registered voters, instead of total population or state citizen population, to apportion the state's legislative districts. Hawaii chose to use registered voters because of the state's large population of military personnel and tourists. Those temporary residents, who were counted in census population but mostly ineligible to vote in Hawaii because of residency requirements, were concentrated primarily on one island of the state and would have skewed voter equality among the state's districts. The Court held that Hawaii's apportionment scheme "satisfie[d] the Equal Protection Clause only because on this record it was found to have produced a distribution of legislators not substantially different from that which would have resulted from the use of a permissible population basis." Burns, 384 U.S. at 93, 86 S.Ct. at 1297. Unfortunately, the Court in Burns did not articulate precisely what constitutes a "permissible population basis."

Kozinski suggests that Burns "can only be explained as an application of the principle of electoral equality." Garza, 918 F.2d at 784. However, this court does not agree that the logic of Burns expresses a preference for electoral equality over representational equality. Although the Burns Court recognized that, under the unique circumstances present in Hawaii, the state was allowed to use an apportionment base other than total population, the Court did not require the state to use an alternative apportionment base, nor did the Court suggest that the use of total population in those circumstances would have been unconstitutional.

The more important lesson that may be gleaned from Burns is that courts should generally defer to the state to chose its own apportionment base, provided that such method yields acceptable results. As the Burns Court stated,

Neither in Reynolds v. Sims nor in any other decision has this Court suggested that the States are required to include aliens, transients, short-term or temporary residents, or persons denied the vote for conviction of crime, in the apportionment base by which their legislators are distributed and against which compliance with the Equal Protection Clause is to be measured. The decision to include or exclude any such group involves choices about the nature of representation with which we have been shown no constitutionally founded reason to interfere. Unless a choice is one the Constitution forbids,10 the resulting apportionment base offends no constitutional bar, and compliance with the rule established in Reynolds v. Sims is to be measured thereby.

384 U.S. at 92, 86 S. Ct. at 1296-97 (footnotes and citations omitted). The Court's general deference to the state's choice of apportionment base is also illustrated by the Court's statement that "Hawaii's special population problems might well have led it to conclude that state citizen population rather than total population should be the basis for comparison." Id. at 94, 86 S.Ct. at 1297 (emphasis added). This permissive language implies that the decision to use an apportionment base other than total population is up to the state, and that courts should not interfere unless the apportionment base is unconstitutionally discriminatory on its face or produces an unacceptably wide variation from total population equality.

In Ellis v. Mayor & City Council of Baltimore, 352 F.2d 123 (4th Cir.1965), a case similar to Burns v. Richardson, this court invalidated an apportionment scheme based on registered voters. In Ellis, we did not foreclose the possibility of basing apportionment on a measure other than total census population, provided that such an alternative resulted in only minor deviations from the census population count. Id. at 129. We did recognize in dictum, however, that total population "is constitutionally unassailable beyond question." Id. at 130.

In sum, this court disagrees with Judge Kozinski's opinion in Garza that the language of the Supreme Court's one person, one vote decisions indicates that electoral equality is more important than representational equality.11 Next, we will examine whether, as a matter of public policy, there is a valid reason to favor electoral equality over representational equality.

The Garza majority defined representational equality only in terms of the right of constituents to have equal access to their elected representatives so that the constituents can effectively convey their concerns and interests to the representatives. See Garza, 918 F.2d at 775. However, the right to petition one's representative is but one facet of the concept of representation.

As the Supreme Court stated in Reynolds v. Sims, "representative government is in essence self-government through the medium of elected representatives of the people, and each and every citizen has an inalienable right to full and effective participation in the political processes of his State's legislative bodies." 377 U.S. at 565, 84 S.Ct. at 1383. Under a representative form of government, people have essentially two fundamental powers with respect to the government.

The first power is the power of an individual to influence what his or her representative does. This power is embodied most obviously by the right to vote--the right to say who one's representative is and the right to change one's representative. The right to vote is reserved only to certain members of society. But people can affect what their representatives do in another way: through their right to petition their representatives to voice their concerns and interests on particular issues. This right is available to everyone, even those who are ineligible to vote. The dichotomy in Garza between electoral equality and representational equality focuses only on the first power of representative government.

The second power in representative government is the power actually exercised by the representative in the governing body, on behalf of his or her constituents. Arguably this power would be nonexistent for constituents if they had no control over their representatives through the right to vote. See Wesberry v. Sanders, 376 U.S. 1, 17, 84 S. Ct. 526, 535, 11 L. Ed. 2d 481 (1964) ("Other rights, even the most basic, are illusory if the right to vote is undermined."). Nevertheless, representatives have an inherent obligation to champion the interests of their constituents. The central power of the governing entity should, in theory, be divided equally among each representative. Although the overall power of the governing body is generally not divisible, each representative individually should have the same ability to influence the actions performed by the governing body as a whole. These representatives should represent roughly the same number of constituents, so that each person, whether or not they are entitled to vote, receives a fair share of the governmental power, through his or her representative. In addition, as Judge Kozinski observed in his opinion in Garza, if representatives can secure governmental benefits and services for their constituents in proportion to their share of the governmental power, then districts of equal population would ensure that all persons receive roughly equal benefits and services from the government. 918 F.2d at 781. Representational equality serves the function of equalizing this second power among all people.

The Garza court's focus only on access to elected representatives is perhaps susceptible to valid criticism.12 However, when all of the aspects of equal representation are considered as a whole, it becomes clear that representational equality is at least as important as electoral equality in a representative democracy.

What, then, should courts do when faced with a situation, such as presented here, where electoral equality and representational equality cannot be achieved simultaneously? This is quintessentially a decision that should be made by the state, not the federal courts, in the inherently political and legislative process of apportionment. See Burns, 384 U.S. at 92, 86 S. Ct. at 1296-97 ("The decision to include or exclude any such group involves choices about the nature of representation with which we have been shown no constitutionally founded reason to interfere.").

In the absence of a clear pronouncement from the Supreme Court on this issue, the district court's actions here should have been tempered by the overriding theme in the Court's prior apportionment cases weighing against judicial involvement. See, e.g., Gaffney v. Cummings, 412 U.S. at 749, 93 S.Ct. at 2329 ("From the very outset, we recognized that the apportionment task, dealing as it must with fundamental 'choices about the nature of representation,' is primarily a political and legislative process.") (quoting Burns, 384 U.S. at 92, 86 S. Ct. at 1296-97). The district court erred in reaching out to extend the federal judiciary's authority in the apportionment process, especially in light of this court's dictum in Ellis v. Mayor & City Council of Baltimore that total population is "constitutionally unassailable beyond question." 352 F.2d 123, 130 (4th Cir.1965). As Defendants contend, North Carolina has specifically chosen total population as its preferred apportionment base,13 and the district court should have respected its choice. By basing its one person, one vote analysis on voting-age population, rather than total population, the district court effectively ignored the state's articulated policy for apportionment.

Even if electoral equality were the paramount concern of the one person, one vote principle, the district court's approach in this action would lead federal courts too far into the "political thicket."14 As the Supreme Court cautioned in Gaffney v. Cummings, federal courts should not "become bogged down in a vast, intractable apportionment slough, particularly when there is little, if anything to be accomplished by doing so." 412 U.S. at 750, 93 S.Ct. at 2330. There is no reason to believe that voting-age population is significantly better than total population in achieving the goal of one person, one vote.

Voting-age population, like total population, is not a perfect indicator of actual voting strength, because it includes persons who are ineligible to vote, such as aliens, temporary residents (e.g., non-resident students, military personnel or other transient persons), and convicted felons. Also, for apportionments based on federal census data, voting-age population, like total population, is inherently inaccurate, because of population shifts during the ten-year period between census counts. See Gaffney, 412 U.S. at 745-48, 93 S.Ct. at 2327-29. Furthermore, using voting-age population as the apportionment base would ignore the voting strength of those persons who are between the ages of 8 and 17 at the time of the apportionment, but who would become eligible to vote before the next apportionment.

Certainly, a state could achieve more precise adherence to electoral equality by basing its apportionment on actual voters and by conducting its apportionment process more frequently than every ten years.15 However, as the Supreme Court has recognized throughout its one person, one vote cases, the process of apportionment must be judged by what is reasonable and practicable. See, e.g., Reynolds v. Sims, 377 U.S. at 577, n. 57, 84 S. Ct. at 1390, n. 57 (" 'We must remember that the machinery of government would not work if it were not allowed a little play in its joints.' ") (quoting Bain Peanut Co. v. Pinson, 282 U.S. 499, 501, 51 S. Ct. 228, 229, 75 L. Ed. 482 (1931)). The Court has consistently held that total population is a reasonable apportionment base.

Finally, Plaintiffs suggest that perhaps a districting plan could be created for Mecklenburg County which would produce de minimis variations in both total population and voting-age population, thereby achieving both electoral equality and representational equality. We will not address this argument because such a proposed plan is not part of the record on appeal. Moreover, as discussed previously, the Supreme Court in Gaffney v. Cummings specifically held that the mere possibility of creating a marginally better districting plan cannot be used to establish a one person, one vote violation. 412 U.S. at 750-51, 93 S. Ct. at 2330-31.

For the foregoing reasons, we conclude that the district court erred in basing its one person, one vote analysis in this action on voting-age population instead of total population. The differences in total population among the districts at issue here are de minimis because the maximum deviation is less than 10%. We therefore vacate the district court's order and remand the case to the district court to receive any additional evidence Plaintiffs may offer to show that the districting plan at issue was the result of bad faith, arbitrariness, or invidious discrimination. If Plaintiffs fail to produce sufficient evidence, the district court shall enter summary judgment in favor of Defendants.

VACATED AND REMANDED WITH INSTRUCTIONS.