concurring.
As an alternative ground for affirmance, the appellees contended at oral argument that the bizarre configuration of New Jersey’s congressional districts is sufficient to demonstrate that the plan was not adopted in “good faith.” This argument, as I understand it, is a claim that the district boundaries are unconstitutional because they are the product of political gerrymandering. Since my vote is decisive in this case, it seems appropriate to explain how this argument influences my analysis of the question that divides the Court. As I have previously pointed out, political gerrymandering is one species of “vote dilution” that is proscribed by the Equal Protection Clause.1 Because an adequate judicial analysis of
*0
*745a gerrymandering claim raises special problems, I shall comment at some length on the legal basis for a gerrymandering claim, the standards for judging such a claim, and their relevance to the present case.
I
Relying on Art. I, §2, of the Constitution, as interpreted in Wesberry v. Sanders, 376 U. S. 1 (1964), and subsequent cases, appellees successfully challenged the congressional districting plan adopted by the New Jersey Legislature. For the reasons stated in Justice Brennan’s opinion for the Court, which I join, the doctrine of stare decisis requires that result. It can be demonstrated, however, that the holding in Wesberry, as well as our holding today, has firmer roots in the Constitution than those provided by Art. I, § 2.
The constitutional mandate contained in Art. I, §2, concerns the number of Representatives that shall be “apportioned among the several States.”2 The section says nothing about the composition of congressional districts within a State.3 Indeed, the text of that section places no restriction whatsoever on the power of any State to define the group of persons within the State who may vote for particular candidates. If a State should divide its registered voters into separate classes defined by the alphabetical order of their initials, by their age, by their period of residence in the State, or even by their political affiliation, such a classification would not be barred by the text of Art. I, § 2, even if the classes contained widely different numbers of voters.
*746As Justice Harlan pointed out in his dissenting opinion in Wesberry, prior to the Civil War the principle of numerical equality of representation was actually contradicted by the text of Art. I, § 2, which provided that the “whole Number of free Persons” should be counted, that certain Indians should be excluded, and that only “three-fifths of all other Persons” should be added to the total.4 In analyzing the Constitution, we cannot ignore the regrettable fact that, as originally framed, it expressly tolerated the institution of slavery. On the other hand, neither can we ignore the basic changes caused by the Civil War Amendments. They planted the roots that firmly support today’s holding.
The abolition of slavery and the guarantees of citizenship and voting rights contained in the Thirteenth, Fourteenth, and Fifteenth Amendments effectively repealed Art. I, §2’s requirement that some votes be given greater weight than others. It remains true, however, that Art. I, §2, does not itself contain any guarantee of equality of representation. The source of that guarantee must be found elsewhere. But as Justice Clark perceptively noted in his partial concurrence *747in Wesberry — and as Justice Black had written earlier in his dissent in Colegrove v. Green, 328 U. S. 549, 569 (1946) — that guarantee is firmly grounded in the Equal Protection Clause of the Fourteenth Amendment.5 Even Justice Harlan’s powerful dissent in Wesberry could find no flaw in that analysis.
In its review of state laws redefining congressional districts subsequent to Wesberry v. Sanders, the Court has not found it necessary to rely on the Equal Protection Clause. That Clause has, however, provided the basis for applying the “one person, one vote” standard to other electoral districts. See, e. g., Baker v. Carr, 369 U. S. 186 (1962); Reynolds v. Sims, 377 U. S. 533 (1964); Avery v. Midland County, 390 U. S. 474 (1968). Even if Art. I, §2, were wholly disregarded, the “one person, one vote” rule would unquestionably apply to action by state officials defining congressional districts just as it does to state action defining state legislative districts.6
*748The Equal Protection Clause requires every State to govern impartially. When a State adopts rules governing its election machinery or defining electoral boundaries, those rules must serve the interests of the entire community. See Reynolds v. Sims, supra, at 565-566. If they serve no purpose other than to favor one segment — whether racial, ethnic, religious, economic, or political — that may occupy a position of strength at a particular point in time, or to disadvantage a politically weak segment of the community, they violate the constitutional guarantee of equal protection.
In Gomillion v. Lightfoot, 364 U. S. 339, 340 (1960), the Court invalidated a change in the city boundaries of Tuskegee, Alabama, “from a square to an uncouth twenty-eight-sided figure” excluding virtually all of the city’s black voters. The Court’s opinion identified the right that had been violated as a group right:
“When a legislature thus singles out a readily isolated segment of a racial minority for special discriminatory treatment, it violates the Fifteenth Amendment. In no case involving unequal weight in voting distribution that has come before the Court did the decision sanction a differentiation on racial lines whereby approval was given to unequivocal withdrawal of the vote solely from colored citizens.” Id., at 346.
Although the Court- explicitly rested its decision on the Fifteenth Amendment, the analysis in Justice Whittaker’s concurring opinion — like Justice Clark’s in Wesberry — is equally coherent, see 364 U. S., at 349. Moreover, the Court has subsequently treated Gomillion as though it had been decided on equal protection grounds. See Whitcomb v. Chavis, 403 U. S. 124, 149 (1971).
*749Gomillion involved complete geographical exclusion of a racially identified group. But in case after case arising under the Equal Protection Clause the Court has suggested that “dilution” of the voting strength of cognizable 'political as well as racial groups may. be unconstitutional. Thus, the question reserved in Fortson v. Dorsey, 379 U. S. 433, 439 (1965), related to an apportionment scheme that might “operate to minimize or cancel out the voting strength of racial or political elements of the voting population.” See also Gaffney v. Cummings, 412 U. S. 735, 751, 754 (1973); White v. Regester, 412 U. S. 755, 765-770 (1973); Whitcomb v. Chavis, supra, at 143-144; Burns v. Richardson, 384 U. S. 73, 88-89 (1966). In his separate opinion in Williams v. Rhodes, 393 U. S. 23, 39 (1968), Justice Douglas pointed out that the Equal Protection Clause protects “voting rights and political groups ... as well as economic units, racial communities, and other entities.” And in Abate v. Mundt, 403 U. S. 182, 187 (1971), the Court noted the absence of any “built-in bias tending to favor particular political interests or geographic areas.” In his dissenting opinion today, Justice White seems to agree that New Jersey’s plan would violate the Equal Protection Clause if it “invidiously discriminated against a racial or political group.” Post, at 783.
There is only one Equal Protection Clause. Since the Clause does not make some groups of citizens more equal than others, see Zobel v. Williams, 457 U. S. 55, 71 (1982) (Brennan, J., concurring), its protection against vote dilution cannot be confined to racial groups. As long as it proscribes gerrymandering against such groups, its proscription must provide comparable protection for other cognizable groups of voters as well. As I have previously written:
“In the line-drawing process, racial, religious, ethnic, and economic gerrymanders are all species of political gerrymanders.
“From the standpoint of the groups of voters that are affected by the line-drawing process, it is also important *750to recognize that it is the group’s interest in gaining or maintaining political power that is at stake. The mere fact that a number of citizens share a common ethnic, racial, or religious background does not create the need for protection against gerrymandering. It is only when their common interests are strong enough to be manifested in political action that the need arises. For the political strength of a group is not a function of its ethnic, racial, or religious composition; rather it is a function of numbers — specifically the number of persons who will vote in the same way.” Mobile v. Bolden, 446 U. S. 55, 88 (1980) (concurring in judgment).
See Cousins v. City Council of Chicago, 466 F. 2d 830, 851-852 (CA7) (Stevens, J., dissenting), cert. denied, 409 U. S. 893 (1972).7
II
Like Justice White, I am convinced that judicial preoccupation with the goal of perfect population equality is an inadequate method of judging the constitutionality of an apportionment plan. I would not hold that an obvious gerrymander is wholly immune from attack simply because it comes closer to perfect population equality than every competing plan. On the other hand, I do not find any virtue in the proposal to relax the standard set forth in Wesberry and subsequent cases, and to ignore population disparities after some arbitrarily defined threshold has been crossed.8 As one com*751mentator has written: “Logic, as well as experience, tells us . . . that there can be no total sanctuaries in the political thicket, else unfairness will simply shift from one form to another.”9 Rather, we should supplement the population equality standard with additional criteria that are no less “judicially manageable. ” In evaluating equal protection challenges to districting plans, just as in resolving such attacks on other forms of discriminatory action, I would consider whether the plan has a significant adverse impact on an identifiable political group, whether the plan has objective indicia of irregularity, and then, whether the State is able to produce convincing evidence that the plan nevertheless serves neutral, legitimate interests of the community as a whole.
Until two decades ago, constrained by its fear of entering a standardless political thicket, the Court simply abstained from any attempt to judge the constitutionality of legislative apportionment plans, even when the districts varied in population from 914,053 to 112,116. See Colegrove v. Green, 328 U. S., at 557. In Baker v. Carr, 369 U. S. 186 (1962), and Reynolds v. Sims, 377 U. S. 533 (1964), the Court abandoned that extreme form of judicial restraint and enunciated the “one person, one vote” principle. That standard is “judicially manageable” because census data are concrete and reasonably reliable and because judges can multiply and divide.
Even as a basis for protecting voters in their individual capacity, the “one person, one vote” approach has its shortcomings. Although population disparities are easily quantified, the standard provides no measure of the significance of any numerical difference. It is easy to recognize the element of *752unfairness in allowing 112,116 voters to elect one Congressman while another is elected by 914,053. But how significant is the difference between census counts of 527,472 and 523,798? Given the birth rate, the mortality rate, the transient character of modern society, and the acknowledged errors in the census, we all know that such differences may vanish between the date of the census and the date of the next election. Absolute population equality is impossible to achieve.
More important, mere numerical equality is not a sufficient guarantee of equal representation. Although it directly protects individuals, it protects groups only indirectly at best. See Reynolds v. Sims, supra, at 561. A voter may challenge an apportionment scheme on the ground that it gives his vote less weight than that of other voters; for that purpose it does not matter whether the plaintiff is combined with or separated from others who might share his group affiliation. It is plainly unrealistic to assume that a smaller numerical disparity will always produce a fairer districting plan. Indeed, as Justice Harlan correctly observed in Wells v. Rockefeller, 394 U. S. 542, 551 (1969), a standard “of absolute equality is perfectly compatible with ‘gerrymandering’ of the worst sort. A computer may grind out district lines which can totally frustrate the popular will on an overwhelming number of critical issues.” Since Justice Harlan wrote, developments in computer technology have made the task of the gerrymanderer even easier. See post, at 776 (WHITE, J., dissenting).10
*753The imperfections in the numerical standard do not, of course, render it useless. It provides one neutral criterion for evaluating a districting plan. Numerical disparities may provide sufficient basis for shifting the burden of justification to the State. Moreover, if all other factors were in equipoise, it would be proper to conclude that the plan that most nearly attains the goal of complete equality would be the fairest plan. The major shortcoming of the numerical standard is its failure to take account of other relevant — indeed, more important — criteria relating to the fairness of group participation in the political process. To that extent, it may indeed be counterproductive. See Gaffney v. Cummings, 412 U. S., at 748-749.11
To a limited extent the Court has taken cognizance of discriminatory treatment of groups of voters. The path the Court has sometimes used to enter this political thicket is marked by the label “intent.” A finding that the majority deliberately sought to make it difficult for a minority group to elect representatives may provide a sufficient basis for holding that an objectively neutral electoral plan is unconstitutional. See Rogers v. Lodge, 458 U. S. 613, 616-617 (1982). For reasons that I have already set forth at length, this standard is inadequate. See id., at 642-650 (Stevens, J., dissenting); Mobile v. Bolden, 446 U. S., at 83 (Stevens, J., concurring in judgment). I would not condemn a legislature’s districting plan in the absence of discriminatory impact simply because its proponents were motivated, in part, by partisanship or group animus. Legislators are, after all, politicians; it is unrealistic to attempt to proscribe all political considerations in the essentially political process of redistricting. In the long run, constitutional adjudication that is premised on a case-by-case appraisal of the subjective intent of local decisionmakers *754cannot possibly satisfy the requirement of impartial administration of the law that is embodied in the Equal Protection Clause of the Fourteenth Amendment. On the other hand, if a plan has a significant adverse impact upon a defined political group, an additional showing that it departs dramatically from neutral criteria should suffice to shift the task of justification to the state defendants.
For a number of reasons, this is a burden that plaintiffs can meet in relatively few cases. As a threshold matter, plaintiffs must show that they are members of an identifiable political group whose voting strength has been diluted. They must first prove that they belong to a politically salient class, see supra, at 749-750, one whose geographical distribution is sufficiently ascertainable that it could have been taken into account in drawing district boundaries.12 Second, they must prove that in the relevant district or districts or in the State as a whole, their proportionate voting influence has been adversely affected by the challenged scheme.13 Third, plain*755tiffs must make a prima facie showing that raises a rebuttable presumption of discrimination.
One standard method by which members of a disadvantaged political group may establish a dilution of their voting rights is by reliance on the “one person, one vote” principle, which depends on a statewide statistical analysis. But prima facie evidence of gerrymandering can surely be presented in other ways. One obvious type of evidence is the shape of the district configurations themselves. One need not use Justice Stewart’s classic definition of obscenity — “I know it when I see it”14 — as an ultimate standard for judging the constitutionality of a gerrymander to recognize that dramatically irregular shapes may have sufficient probative force to call for an explanation.15
Substantial divergences from a mathematical standard of compactness may be symptoms of illegitimate gerrymandering. As Dr. Ernest Reock, Jr., of Rutgers University has written: “Without some requirement of compactness, the boundaries of a district may twist and wind their way across the map in fantastic fashion in order to absorb scattered *756pockets of partisan support.”16 To some extent, geographical compactness serves independent values; it facilitates political organization, electoral campaigning, and constituent representation.17 A number of state statutes and Constitutions require districts to be compact and contiguous. These standards have been of limited utility because they have not been defined and applied with rigor and precision.18 Yet Professor Reock and other scholars have set forth a number of methods of measuring compactness that can be computed with virtually the same degree of precision as a population count.19 It is true, of course, that the significance of a par*757ticular compactness measure may be difficult to evaluate, but as the figures in this case demonstrate, the same may be said of population disparities. In addition, although some deviations from compactness may be inescapable because of the geographical configuration or uneven population density of a particular State,20 the relative degrees of compactness of dif*758ferent district maps can always be compared. As with the numerical standard, it seems fair to conclude that drastic departures from compactness are a signal that something may be amiss.
Extensive deviation from established political boundaries is another possible basis for a prima facie showing of gerrymandering. As we wrote in Reynolds v. Sims: “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.” 377 U. S., at 578-579.21 Subdivision boundaries tend to remain stable over time. Residents of political units such as townships, cities, and counties often develop a community of interest, particularly when the subdivision plays an important role in the provision of governmental services. In addition, legislative districts that do not cross subdivision boundaries are administratively convenient and less likely to confuse the voters.22 Although the significance of deviations from sub*759division boundaries will vary with the number of legislative seats and the number, size, and shape of the State’s subdivisions, the number can be counted23 and alternative plans can be compared.
A procedural standard, although obviously less precise, may also be enlightening. If the process for formulating and adopting a plan excluded divergent viewpoints, openly reflected the use of partisan criteria, and provided no explanation of the reasons for selecting one plan over another, it would seem appropriate to conclude that an adversely affected plaintiff group is entitled to have the majority explain its action.24 On the other hand, if neutral decisionmakers developed the plan on the basis of neutral criteria, if there was an adequate opportunity for the presentation and consideration of differing points of view, and if the guidelines used in selecting a plan were explained, a strong presumption of validity should attach to whatever plan such a process produced.
Although a scheme in fact worsens the voting position of a particular group,25 and though its geographic configuration or *760genesis is sufficiently irregular to violate one or more of the criteria just discussed, it will nevertheless be constitutionally valid if the State can demonstrate that the plan as a whole embodies acceptable, neutral objectives. The same kinds of justification that the Court accepts as legitimate in the context of population disparities would also be available whenever the criteria of shape, compactness, political boundaries, or decisionmaking procedures have sent up warning flags. In order to overcome a prima facie case of invalidity, the State may adduce “legitimate considerations incident to the effectuation of a rational state policy,” Reynolds v. Sims, 377 U. S., at 579, and may also
“show with some specificity that a particular objective requires the specific deviations in its plan, rather than simply relying on general assertions. The showing . . . is flexible, depending on the size of the deviations, the importance of the State’s interests, the consistency with which the plan as a whole reflects those interests, and the availability of alternatives that might substantially vindicate those interests yet approximate population equality more closely.” Ante, at 741.26
If a State is unable to respond to a plaintiff’s prima facie case by showing that its plan is supported by adequate neutral criteria, I believe a court could properly conclude that the challenged scheme is either totally irrational or entirely *761motivated by a desire to curtail the political strength of the affected political group. This does not mean that federal courts should invalidate or even review every apportionment plan that may have been affected to some extent by partisan legislative maneuvering.27 But I am convinced that the Judiciary is not powerless to provide a constitutional remedy in egregious cases.28
Ill
In this case it is not necessary to go beyond the reasoning in the Court’s opinions in Wesberry v. Sanders, 376 U. S. 1 (1964), Kirkpatrick v. Preisler, 394 U. S. 526 (1969), and *762White v. Weiser, 412 U. S. 783 (1973), to reach the correct result. None of the additional criteria that I have mentioned would cast any doubt on the propriety of the Court’s holding in this case. Although I need not decide whether the plan’s shortcomings regarding shape and compactness, subdivision boundaries, and neutral decisionmaking would establish a prima facie case, these factors certainly strengthen my conclusion that the New Jersey plan violates the Equal Protection Clause.
A glance at the map, ante, following p. 744, shows district configurations well deserving the kind of descriptive adjectives — “uncouth”29 and “bizarre”30 — that have traditionally been used to describe acknowledged gerrymanders. I have not applied the mathematical measures of compactness to the New Jersey map, but I think it likely that the plan would not fare well. In addition, while disregarding geographical compactness, the redistricting scheme wantonly disregards county boundaries. For example, in the words of a commentator: “In a flight of cartographic fancy, the Legislature packed North Jersey Republicans into a new district many call ‘the Swan.’ Its long neck and twisted body stretch from the New York suburbs to the rural upper reaches of the Delaware River.” That district, the Fifth, contains segments of at least seven counties. The same commentator described the Seventh District, comprised of parts of five counties, as tracing “a curving partisan path through industrial Elizabeth, liberal, academic Princeton and largely Jewish Marl*763boro in Monmouth County. The resulting monstrosity was called ‘the Fishhook’ by detractors.” 40 Congressional Quarterly 1193-1195 (1982).31
Such a map prompts an inquiry into the process that led to its adoption. The plan was sponsored by the leadership in the Democratic Party, which controlled both houses of the state legislature as well as the Governor’s office, and was signed into law the day before the inauguration of a Republican Governor. The legislators never formally explained the guidelines used in formulating their plan or in selecting it over other available plans. Several of the rejected plans contained districts that were more nearly equal in population, more compact, and more consistent with subdivision boundaries, including one submitted by a recognized expert, Dr. Ernest Reock, Jr., whose impartiality and academic credentials were not challenged. The District Court found that the Reock Plan “was rejected because it did not reflect the leadership’s partisan concerns.” Daggett v. Kimmelman, 535 F. Supp. 978, 982 (NJ 1982). This conclusion, which arises naturally from the absence of persuasive justifications for the rejection of the Reock Plan, is buttressed by a letter written to Dr. Reock by the Democratic Speaker of the New Jersey General Assembly. This letter frankly explained the importance to the Democrats of taking advantage of their opportunity to control redistricting after the 1980 census. The Speaker justified his own overt partisanship by describing the political considerations that had motivated the Republican majority in the adoption of district plans in New *764Jersey in the past — and in other States at the present.32 In sum, the record indicates that the decisionmaking process leading to adoption of the challenged plan was far from neutral. It was designed to increase the number of Democrats, and to decrease the number of Republicans, that New Jersey’s voters would send to Congress in future years.33 Finally, the record does not show any legitimate justifications for the irregularities in the New Jersey plan, although concededly the case was tried on a different theory in the District Court.
Because I have not made a comparative study of other dis-tricting plans, and because the State has not had the opportu*765nity to offer justifications specifically directed toward the additional concerns I have discussed, I cannot conclude with absolute certainty that the New Jersey plan was an unconstitutional partisan gerrymander. But I am in full agreement with the Court’s holding that, because the plan embodies deviations from population equality that have not been justified by any neutral state objective, it cannot stand. Further, if population equality provides the only check on political gerrymandering, it would be virtually impossible to fashion a fair and effective remedy in a case like this. For if the shape of legislative districts is entirely unconstrained, the dominant majority could no doubt respond to an unfavorable judgment by providing an even more grotesque-appearing map that reflects acceptable numerical equality with even greater political inequality. If federal judges can prevent that consequence by taking a hard look at the shape of things to come in the remedy hearing, I believe they can also scrutinize the original map with sufficient care to determine whether distortions have any rational basis in neutral criteria. Otherwise, the promise of Baker v. Carr and Reynolds v. Sims— that judicially manageable standards can assure “[f]ull and effective participation by all citizens,” 377 U. S., at 565 — may never be fulfilled.
See Cousins v. City Council of Chicago, 466 F. 2d 830, 848-853 (CA7) (Stevens, J., dissenting), cert. denied, 409 U. S. 893 (1972); Mobile v. Bolden, 446 U. S. 55, 86-89 (1980) (Stevens, J., concurring in judgment); Rogers v. Lodge, 458 U. S. 613, 652 (1982) (Stevens, J., dissenting).
Article I, §2, provides, in part:
“Representatives and direct Taxes shall be apportioned among the several States which may be included within this Union, according to their respective Numbers, which shall be determined by adding to the whole Number of free Persons, including those bound to Service for a Term of Years, and excluding Indians not taxed, three fifths of all other Persons.” U. S. Const., Art. I, §2, cl. 3 (emphasis supplied).
During the first 50 years of our Nation’s history, it was a widespread practice to elect Members of the House of Representatives as a group on a statewide basis. Wesberry v. Sanders, 376 U. S. 1, 8 (1964).
“Representatives were to be apportioned among the States on the basis of free population plus three-fifths of the slave population. Since no slave voted, the inclusion of three-fifths of their number in the basis of apportionment gave the favored States representation far in excess of their voting population. If, then, slaves were intended to be without representation, Article I did exactly what the Court now says it prohibited: it ‘weighted’ the vote of voters in the slave States. Alternatively, it might have been thought that Representatives elected by free men of a State would speak also for the slaves. But since the slaves added to the representation only of their own State, Representatives from the slave States could have been thought to speak only for the slaves of their own States, indicating both that the Convention believed it possible for a Representative elected by one group to speak for another nonvoting group and that Representatives were in large degree still thought of as speaking for the whole population of a State.” Id., at 27-28.
Reading a “one person, one vote” requirement into Art. I, § 2, is historically as well as textually unsound. See Kelly, Clio and the Court: An Illicit Love Affair, 1965 S. Ct. Rev. 119, 135-136.
That Clause “does not permit the States to pick out certain qualified citizens or groups of citizens and deny them the right to vote at all.... No one would deny that the equal protection clause would also prohibit a law that would expressly give certain citizens a half-vote and others a full vote. The probable effect of the 1901 State Apportionment Act in the coming election will be that certain citizens, and among them the appellants, will in some instances have votes only one-ninth as effective in choosing representatives to Congress as the votes of other citizens. Such discriminatory legislation seems to me exactly the kind that the equal protection clause was intended to prohibit.” Colegrove v. Green, 328 U. S., at 569 (Black, J., dissenting), quoted in part in Wesberry v. Sanders, supra, at 19 (Clark, J., concurring in part and dissenting in part).
The “one person, one vote” rule, like the Equal Protection Clause in which it is firmly grounded, provides protection against more than one form of discrimination. In the cases in which the rule was first developed, district boundaries accorded significantly less weight to individual votes in the most populous districts. But it was also clear that those boundaries maximized the political strength of rural voters and diluted the political power of urban voters. See A. Hacker, Congressional Districting: The Issue of Equal Representation 20-26 (1963); see generally Standards for Congressional Districts (Apportionment), Hearings before Subcommittee No. 2 of the House Committee on the Judiciary on H. R. 73, H. R. 575, *748H. R. 8266, and H. R. 8473, 86th Cong., 1st Sess., 65-90 (1959). The primary consequence of the rule has been its protection of the individual voter, but it has also provided one mechanism for identifying and curtailing discrimination against cognizable groups of voters.
Similarly, the motivation for the gerrymander turns on the political strength of members of the group, derived from cohesive voting patterns, rather than on the source of their common interests. 466 F. 2d, at 852.
The former would appear to be consistent with what the Court has written in this case, ante, at 734-735, n. 6; the latter would be consistent with what Justice White has written in dissent, post, at 780-783. Either of these approaches would leave the door to unrestricted gerrymandering wide open. See Engstrom, The Supreme Court and Equipopulous Gerrymandering: A Remaining Obstacle in the Quest for Pair and Effective Representation, 1976 Ariz. State L. J. 277, 285-286, 296; Baker, Quantita*751tive and Descriptive Guidelines to Minimize Gerrymandering, 219 Annals N. Y. Acad. Sci. 200, 208 (1973) (“If more specific guidelines to minimize gerrymandering are not forthcoming, then a great democratic principle— one man, one vote — will have degenerated into a simplistic arithmetical facade for discriminatory cartography on an extensive scale”).
Dixon, The Court, the People, and “One Man, One Vote,” in Reapportionment in the 1970s, p. 32 (N. Polsby ed. 1971).
Computers now make it possible to generate a large number of alternative plans, consistent with equal population guidelines and various other criteria, in a relatively short period of time, and to analyze the political characteristics of each one in considerable detail. In contrast, “[i]n the 1970's round of reapportionment, some states were barely able to generate a single reapportionment plan in the time allotted to the task.” National Conference of State Legislatures, Reapportionment: Law and Technology 55 (June 1980); see also Engstrom, supra n. 8, at 281-282.
See Edwards, The Gerrymander and “One Man, One Vote,” 46 N. Y. U. L. Rev. 879 (1971); Elliott, Prometheus, Proteus, Pandora, and Procrustes Unbound: The Political Consequences of Reapportionment, 37 U. Chi. L. Rev. 474, 483-488 (1970); Engstrom, supra n. 8.
Identifiable groups will generally be based on political affiliation, race, ethnic group, national origin, religion, or economic status, but other characteristics may become politically significant in a particular context. See Clinton, Further Explorations in the Political Thicket: The Gerrymander and the Constitution, 59 Iowa L. Rev. 1, 38-39 (1973) (cognizable interest group with coherent and identifiable legislative policy); Comment, Political Gerrymandering: A Statutory Compactness Standard as an Antidote for Judicial Impotence, 41 U. Chi. L. Rev. 398, 407-408 (1974) (clearly identifiable and stable group).
The difficulty in making this showing stems from the existence of alternative strategies of vote dilution. Depending on the circumstances, vote dilution may be demonstrated if a population concentration of group members has been fragmented among districts, or if members of the group have been overconcentrated in a single district greatly in excess of the percentage needed to elect a candidate of their choice. See Mobile v. Bolden, 446 U. S., at 91, and n. 13 (Stevens, J., concurring in judgment); Hacker, supra n. 6, at 46-50; cf. Note, Compensatory Racial Reapportionment, 25 Stan. L. Rev. 84, 97-100 (1972) (pointing to the shortcomings of several tests of political strength, including opportunity to cast swing votes and opportunity to elect a representative of their own group).
In litigation under the Voting Rights Act, federal courts have developed some familiarity with the problems of identifying and measuring dilution of *755racial group voting strength. Some of the concepts developed for statutory purposes might be applied in adjudicating constitutional claims by other types of political groups. The threshold showing of harm may be more difficult for adherents of a political party than for members of a racial group, however, because there are a number of possible base-line measures for a party’s strength, including voter registration and past vote-getting performance in one or more election contests. See generally Backstrom, Robins, & Eller, Issues in Gerrymandering: An Exploratory Measure of Partisan Gerrymandering Applied to Minnesota, 62 Minn. L. Rev. 1121, 1131-1189 (1978).
Jacobellis v. Ohio, 378 U. S. 184, 197 (1964).
Professor Dixon quite properly warns against defining gerrymandering in terms of odd shapes. See R. Dixon, Democratic Representation: Reapportionment in Law and Politics 459-460 (1968). At the same time, however, he recognizes that a rule of compactness and contiguity, “if used merely to force an explanation for odd-shaped districts, can have much merit.” Id., at 460. See L. Tribe, American Constitutional Law 760 (1978) (oddity of district’s shape, coupled with racial distribution of the population, should shift the burden of justification to the State).
Reock, Measuring Compactness as a Requirement of Legislative Apportionment, 5 Midwest J. Pol. Sci. 70, 71 (1961). Cf. Backstrom, Robins, & Eller, supra n. 13, at 1126, 1137 (compactness standard cannot eliminate gerrymandering but may reduce the band of discretion available to those drawing district boundaries). It is of course possible to dilute a group’s voting strength even if all districts are relatively compact. Engstrom, supra n. 8, at 280.
See Taylor, A New Shape Measure for Evaluating Electoral District Patterns, 67 Am. Pol. Sci. Rev. 947, 948 (1973). Compactness is not to be confused with physical area. As we stated in Reynolds v. Sims, 377 U. S. 533, 580 (1964): “Modern developments and improvements in transportation and communications make rather hollow, in the mid-1960’s, most claims that deviations from population-based representation can validly be based solely on geographical considerations. Arguments for allowing such deviations in order to insure effective representation for sparsely settled areas and to prevent legislative districts from becoming so large that the availability of access of citizens to their representatives is impaired are today, for the most part, unconvincing.” Nevertheless, although low population density may require geographically extensive districts, different questions are presented by the creation of districts with distorted shapes and irregular, indented boundaries.
One state statute and 21 State Constitutions explicitly require that districts be compact; two state statutes and 27 Constitutions explicitly provide that districts be formed of contiguous territory. See Congressional Research Service, State Constitutional and Statutory Provisions Concerning Congressional and State Legislative Redistricting (June 1981). But see Clinton, supra n. 12, at 2 (ineffective enforcement); Comment, supra n. 12, at 412-413.
The scholarly literature suggests a number of different mathematical measures of compactness, each focusing on different variables. One rela*757tively simple method is to measure the relationship between the area of the district and the area of the smallest possible circumscribing circle. See Reock, supra n. 16, at 71. This calculation is particularly sensitive to the degree of elongation of a given shape. Another simple method is to determine the ratio of a figure’s perimeter to the circumference of the smallest possible circumscribing circle, a measurement that is well suited to measuring the degree of indentation. See Schwartzberg, Reapportionment, Gerrymanders, and the Notion of “Compactness,” 50 Minn. L. Rev. 443-452 (1966). Other measures of compactness are based on the aggregate of the distances from the district’s geometrical or population-weighted center of gravity to each of its points, see Kaiser, An Objective Method for Establishing Legislative Districts, 10 Midwest J. Pol. Sci. 200-223 (1966); Weaver & Hess, A Procedure for Nonpartisan Districting: Development of Computer Techniques, 73 Yale L. J. 288, 296-300 (1963); the degree of indentation of the boundaries of a nonconvex district, see Taylor, supra n. 17; the aggregate length of district boundaries, see Common Cause, Toward a System of “Fair and Effective Representation” 54-55 (1977); Adams, Statute: A Model State Apportionment Process: The Continuing Quest for “Fair and Effective Representation,” 14 Harv. J. Legis. 825, 875-876, and n. 184 (1977); Edwards, supra n. 11, at 894; Walker, One Man-One Vote: In Pursuit Of an Elusive Ideal, 3 Hastings Const. L. Q. 453, 475 (1976); and the ratio of the maximum to the minimum diameters in a district, R. Morrill, Political Redistricting and Geographic Theory 22 (1981). In each case, the smaller the measurement, the more compact the district or districts. See also 1980 Iowa Acts, ch. 1021, § 4b(3)c (setting forth alternative geometrical tests for determining relative compactness of alternative districting plans: the absolute value of the difference between the length and width of the district, and the “ratio of the dispersion of population about the population center of the district to the dispersion of population about the geographic center of the district”).
If a State’s political subdivisions have oddly shaped boundaries, adhering to these boundaries may detract from geographical compactness. See Colo. Rev. Stat. §§2-2-105, 2-2-203 (1980) (legislative explanations that variations from compactness were caused by “the shape of county boundary lines, census enumeration lines, natural boundaries, population den*758sity, and the need to retain compactness of adjacent districts”); Adams, supra n. 19, at 875-876, n. 184.
In addition, geographic compactness may differ from sociopolitical compactness. Baker, supra n. 8, at 205. As one geographer has noted:
“In many regions, the population is uneven, perhaps strung out along roads or railroads. Travel may be easier and cheaper in some directions than in others, such that an elongated district astride a major transport corridor might in fact be the most compact in the sense of minimum travel time for a representative to travel around the district. If so, then a modified criterion, the ratio of the maximum to the minimum travel time, would be a preferred measure.” Morrill, supra n. 19, at 22.-
In Kirkpatrick v. Preisler, 894 U. S. 526, 534, n. 4 (1969), the Court correctly noted that adherence to subdivision boundaries could not prevent gerrymandering. But there it was concerned with the State’s attempt to justify population disparities by a policy of adhering to existing subdivision boundaries. My discussion here is directed toward partisan gerrymandering in a scheme with relatively equipopulous districts. To the extent that dicta in Kirkpatrick reject the notion that respecting subdivision boundaries will not inhibit gerrymandering, I respectfully disagree. See n. 26, infra.
Morrill, supra n. 19, at 25.
See, e. g., Mahan v. Howell, 410 U. S. 315, 319, 323 (1973); Backstrom, Robins, & Eller, supra n. 13, at 1145, n. 71; Morrill, supra n. 19, at 25. The smaller the population of a subdivision relative to the average district population, the more dubious it is to divide it among two or more districts. It is also particularly suspect to divide a particular political subdivision among more than two districts which also contain territory in other subdivisions.
See, e. g., Wright v. Rockefeller, 376 U. S. 52, 73-74 (1964) (Goldberg, J., dissenting); Edwards, supra n. 11, at 881 (the 1961 New York congressional redistricting plan was drawn up by majority party members of a legislative committee and staff without participation by any member of the opposition party; no public hearings were held; the plan was released to the public the day before its adoption; it was approved by a straight party-line vote in a single afternoon at an extraordinary session of the legislature; and the Governor signed the bill the same day).
The State may defend on the grounds that this element has not been adequately shown. For example, if the plaintiffs’ challenge is based on a particular district or districts, the State may be able to show that the *760group’s voting strength is not diluted in the State as a whole. Even if the group’s voting strength has in fact been reduced, the previous plan may have been gerrymandered in its favor. See generally Backstrom, Robins, & Eller, supra n. 13, at 1134-1137 (discussing possible standards of “fair representation”).
In determining whether the State has carried its burden of justification, I would give greater weight to the importance of the State’s interests and the consistency with which those interests are served than to the size of the deviations. Thus I do not share the perspective implied in the Court’s discussion of purported justifications in Kirkpatrick v. Preisler, 394 U. S., at 533-536.
Given the large number of potentially affected political groups, even a neutral, justifiable plan may well change the position of some groups for the worse. In addition, some “vote dilution” will inevitably result from residential patterns; see Backstrom, Robins, & Eller, supra n. 13, at 1127. Although the State may of course adduce this factor in defense of its plan, the criteria for a prima facie case should be demanding enough that they are not satisfied in the case of every apportionment plan. See Mobile v. Bolden, 446 U. S., at 90 (Stevens, J., concurring in judgment) (“the standard cannot condemn every adverse impact on one or more political groups without spawning more dilution litigation than the judiciary can manage”); id., at 93, n. 15 (quoting opinion of Justice Frankfurter in Baker v. Carr, 369 U. S. 186, 267 (1962)).
See Gomillion v. Lightfoot, 364 U. S. 339, 341 (1960) (noting that allegations would “abundantly establish that Act 140 was not an ordinary geographic redistricting measure even within familiar abuses of gerrymandering”). If the Tuskegee map in Gomillion had excluded virtually all Republicans rather than blacks from the city limits, the Constitution would also have been violated. Professor Tribe gives a comparably egregious numerical hypothetical:
“For example, if a jurisdiction consisting of 540 Republicans and 460 Democrats were subdivided randomly into 10 districts, Republicans would probably be elected in six or more districts. However, if malevolent Democrats could draw district lines with precision, they might be able to isolate 100 Republicans in one district and win all the other district elections by a margin of one or two votes, thus capturing 90% of the state legislature while commanding only 46% of the popular vote.” Tribe, supra n. 15, at 756, n. 2.
See Hacker, supra n. 6, at 47-50.
Gomillion v. Lightfoot, supra, at 339.
Indeed, this very map was so described in a recent article entitled New Jersey Map Imaginative Gerrymander, appearing in the Congressional Quarterly: “New Jersey’s new congressional map is a four-star gerrymander that boasts some of the most bizarrely shaped districts to be found in the nation.” 40 Congressional Quarterly 1190 (1982). A quick glance at congressional districting maps for the other 49 States lends credence to this conclusion. See 1983-1984 Official Congressional Directory 989-1039 (1983).
The same commentator described the Thirteenth District in this manner: “In an effort to create a ‘dumping ground’ for Republican votes troubling to Democrats Hughes and Howard, the Legislature established a 13th District that stretches all over the map, from the Philadelphia suburbs in Camden County to the New York suburbs in Monmouth County.” 40 Congressional Quarterly, at 1198. At oral argument, we observed the likeness between the boundaries of yet another district — the Fourth — and the shape of a running back. Tr. of Oral Arg. 21.
“Congressional redistricting in New Jersey must also be viewed from the more broad-based national perspective. The Republican party is only 27 votes short of absolute control of Congress. With a shift of population and consequently Congressional seats from the traditionally Democratic urban industrial states to the more Republican dominated sun-belt states the redistricting process is viewed by Republicans as an opportunity to close that 27 vote margin, or perhaps even overcome it entirely.” 535 F. Supp., at 991.
Copies of the letter were sent to all Democratic legislators.
Although Circuit Judge Gibbons disagreed with the holding of the District Court in this case, the concluding paragraphs of his dissenting opinion unambiguously imply that he would have no difficulty identifying this as a case in which the district lines were drawn in order to disadvantage an identifiable political group. He wrote:
“The apportionment map produced by P. L. 1982, c.l leaves me, as a citizen of New Jersey, disturbed. It creates several districts which are anything but compact, and at least one district which is contiguous only for yachtsmen. While municipal boundaries have been maintained, there has been little effort to create districts having a community of interests. In some districts, for example, different television and radio stations, different newspapers, and different transportation systems serve the northern and southern localities. Moreover the harshly partisan tone of Speaker Christopher Jackman’s letter to Ernest C. Reock, Jr. is disedifying, to say the least. It is plain, as well, that partisanship produced artificial bulges or appendages of two districts so as to place the residences of Congressmen Smith and Courier in districts where they would be running against incumbents.” Id., at 984.