dissenting.
Before the Court is an application seeking to stay an order of a three-judge District Court pending final disposition of an appeal to this Court under 28 U. S. C. § 1253. The challenged order directs the State of New Jersey to conduct upcoming elections for Members of the House of Representatives pursuant to a reapportionment plan adopted by the District Court as a remedy for the constitutional violation found in New Jersey’s 1982 reapportionment plan. See Karcher v. Daggett, 462 U. S. 725 (1983), aff’g 535 F. Supp. 978 (NJ 1982). Because I believe that the District Court has acted beyond the scope of its authority in correcting the relevant constitutional violation, I would grant the application for stay and remand the case to the District Court for implementation of an alternative plan. I therefore dissent.
I
Following the 1980 decennial census, the State of New Jersey was required to decrease its membership in the United States House of Representatives from 15 to 14. To satisfy this requirement, the State enacted a congressional reapportionment scheme in January 1982 (hereinafter referred to as the Feldman Plan) that eliminated one of the State’s congressional districts and substantially changed the geographical boundaries used to define the re-*912malning districts. Although the Feldman Plan contained the requisite 14 districts, it suffered from significant numerical inequalities in population between each of those districts. In particular, given an “ideal” district population of 526,059, the average deviation from the norm was 0.1384%, or 726 people. Moreover, the difference between the largest district and the smallest district was 3,674 people, or 0.6984% of the average district.
The Feldman Plan was challenged by several interested parties and, primarily because of these population variances, was declared unconstitutional by the District Court. Daggett v. Kimmelman, 535 F. Supp. 978 (NJ 1982). That order was stayed pending appeal to this Court, Karcher v. Daggett, 455 U. S. 1303 (1982) (Brennan, J., in chambers), leaving the plan in effect during the 1982 congressional elections. We noted probable jurisdiction, 457 U. S. 1131 (1982), and subsequently affirmed the decision and order of the District Court, Karcher v. Daggett, 462 U. S. 725 (1983).
In Karcher, this Court reaffirmed that Art. I, §2, of the Constitution “ ‘permits only the limited population variances which are unavoidable despite a good-faith effort to achieve absolute equality, or for which justification is shown.”’ Id., at 730 (quoting Kirkpatrick v. Preisler, 394 U. S. 526, 531 (1969)); see White v. Weiser, 412 U. S. 783, 790 (1973). Applying that standard to the Feldman Plan, we concluded that the numerical variances described above, when combined with evidence that alternative plans available to the State contained smaller maximum deviations, demonstrated that New Jersey had not come “as nearly as practicable” to population equality among districts. 462 U. S., at 730, 739-740; Wesberry v. Sanders, 376 U. S. 1, 7-8, 18 (1964). Nor was the State able to prove that each significant variance among its districts was necessary to achieve some legitimate state objective. 462 U. S., at 740-744. We therefore concluded that the population inequalities existing under the Feldman Plan were both constitutionally significant and unjustified by legitimate state goals. Accordingly, we affirmed the District Court’s holding that the plan was unconstitutional.
On remand from our decision, the three-judge District Court allowed the State until February 3, 1984, to enact another redistricting plan that would meet constitutional requirements. Although the state legislature adopted an alternative plan (Senate *913Bill 3564, hereinafter referred to as the Senate Plan), it was vetoed by the Governor and had insufficient support for reenactment over that veto. Given this failure by the State’s political process, the District Court convened a hearing on February 7, 1984, to choose a proper remedy for the uncorrected constitutional violation. See Daggett v. Kimmelman, 580 F. Supp. 1259 (NJ 1984). At that hearing, all parties agreed that the court should select a redistricting plan from among several alternatives offered by interested parties rather than allow the upcoming congressional elections to proceed on an at-large basis. Id., at 1261. Cf. 2 U. S. C. § 2a(c).
At least six separate redistricting proposals were advanced by various parties before the District Court. Most important for present purposes were the Senate Plan and a plan submitted by various Republican congressional candidates who were the original plaintiffs in this litigation (hereinafter referred to as the Forsythe Plan).1 In its discussion of the Senate Plan, the court first noted that its districts “are virtually identical” to those included in the unconstitutional Feldman Plan. Slight geographical changes had been made, however, resulting in an average deviation from the ideal district of less than 12 people and a maximum variation between the largest and smallest districts of only 67 people. Despite this apparent success in eliminating numerical inequalities, the court refused to accept the Senate Plan as a remedy for the constitutional violation we found last Term in Karcher v. Daggett, supra. In particular, the court found that the plan not only failed to “achieve as small an overall or mean deviation as other plans,” but also retained the “most glaring defects in the Feldman Plan,” including “an obvious absence of compactness, and an intentional gerrymander in favor of certain Democratic representatives.” 580 F. Supp., at 1262.
*914In contrast, the Forsythe Plan produced a maximum variation of only 25 people. And, although the plan required the splitting of two municipalities, its “two great advantages . . . over any of the others, are the achievement of smaller population deviations, and the creation of more compact districts.” Id., at 1264. Thus, the court concluded, the Forsythe Plan “most nearly fits the appropriate criteria for a court considering a congressional reapportionment plan as a remedy for an unconstitutional reapportionment statute.” Id., at 1264-1265. See infra, this page and 915.
In sum, the District Court acknowledged that each of the submitted plans improved substantially on the numerical disparities that led us to conclude last Term that the Feldman Plan was unconstitutional. Nonetheless, the District Court chose the Forsythe Plan as the appropriate replacement for the unconstitutional Feldman Plan because it “creat[ed] more compact districts” than the Senate Plan and there was “no evidence” that “it is designed to achieve partisan advantage.” 580 F. Supp., at 1264.
II
Before choosing among these alternative plans, the District Court explicitly stated the legal principles that it believed should control its remedial decision. The court first summarized the constitutional standard reaffirmed by this Court in Karcher v. Daggett, see supra, at 912, and then specifically noted that our opinion in Karcher had “declin[ed] to rely, as a constitutional violation, on the obviously partisan purposes behind the Feldman Plan.” 580 F. Supp., at 1261. The court nonetheless refused to limit its analysis to the numerical inequalities that triggered our constitutional holding; instead, the court examined the alternative plans using the following principle:
“While Karcher v. Daggett considers what interests may be taken into account by state legislatures in justifying deviations from the ideal of district population equality based on the decennial census, it also provides useful instruction to district courts faced, as we are, with selecting a districting plan because of a failure in the legislative process. We may take into account at least those factors which the Court has recognized as legitimate, namely: making districts compact, preserving municipal boundaries, preserving cores of prior districts, avoiding contests between incumbents, and inhibit*915ing gerrymandering. With those factors in mind we turn to the several plans which have been proposed.” Id., at 1261-1262.
In my view, the District Court’s responsibility in remedying the constitutional violation we found last Term in Karcher does not reach that far. Although two Justices wrote separately to note that the political gerrymandering evident from the geographical boundaries included in the Feldman Plan might be worthy of constitutional challenge, see 462 U. S., at 744 (Stevens, J., concurring); id., at 784 (Powell, J., dissenting), the Court’s finding of a constitutional violation was premised exclusively on numerical inequalities between congressional districts. Once these population disparities are eliminated, our prior cases make clear that the District Court is charged with selecting an alternative plan that accords deference to the policies and preferences that have been expressed previously by the State.
This was precisely the situation presented by White v. Weiser, 412 U. S. 783 (1973). After finding that the State’s reapportionment plan was unconstitutional because of significant, yet avoidable, population variances between districts, the District Court had to choose among several remedial plans proffered by interested parties. More specifically,
“[t]he District Court properly rejected S. B. 1 [the unconstitutional state plan], but it had before it both Plan B and Plan C, and there remains the question whether the court correctly chose to implement the latter. Plan B adhered to the basic district configurations found in S. B. 1, but adjusted the district lines, where necessary, in order to achieve maximum population equality among districts. Each district in Plan B contained generally the same counties as the equivalent district in S. B. 1. Plan C, on the other hand, was based entirely upon population considerations and made no attempt to adhere to the district configurations found in S. B. 1. . . . After deciding that S. B. 1 was unacceptable, the District Court ordered the implementation of Plan C.” Id., at 793-794 (footnotes omitted).
Although the appellees in White v. Weiser defended the lower court’s selection of Plan C because it was “significantly more compact and contiguous than either S. B. 1 or Plan B” and because its selection was “an exercise of the remedial discretion of the District Court,” id., at 794, we rejected the lower court’s remedial *916choice and remanded for further proceedings. When fashioning a constitutional remedy in this context, we explained, the District Court must defer to any state policies that are “consistent with constitutional norms and . . . not [themselves] vulnerable to legal challenge.” Id., at 797.
“From the beginning, we have recognized that ‘reapportionment is primarily a matter for legislative consideration and determination, and that judicial relief becomes appropriate only when a legislature fails to reapportion according to federal constitutional requisites in a timely fashion after having had an adequate opportunity to do so.’ ... We have adhered to the view that state legislatures have ‘primary jurisdiction’ over legislative reapportionment. . . . Just as a federal district court, in the context of legislative reapportionment, should follow the policies and preferences of the State, as expressed in statutory and constitutional provisions or in the reapportionment plans proposed by the state legislature, whenever adherence to state policy does not detract from the requirements of the Federal Constitution, we hold that a district court should similarly honor state policies in the context of congressional reapportionment. In fashioning a reapportionment plan or in choosing among plans, a district court should not pre-empt the legislative task nor ‘intrude upon state policy any more than necessary.’” Id., at 794-795 (citations omitted).
See also Upham v. Seamon, 456 U. S. 37, 41-43 (1982) (per curiam).
Pursuant to these standards, the Court in White v. Weiser reversed the lower court’s selection of a reapportionment plan, and remanded for the imposition of another plan. This was necessary because the District Court had not chosen that plan which, while eliminating the constitutional violation, would be most in accord with the State’s policy preferences. See 412 U. S., at 796 (the District Court “should have implemented [the plan] which most clearly approximated the reapportionment plan of the state legislature, while satisfying constitutional requirements”); Upham v. Seamon, supra, at 42-43. Significantly, it was irrelevant to our analysis that the last state plan formally adopted to implement those policy judgments had itself been declared unconstitutional. See 412 U. S., at 796 (“even if the districts in [the plan chosen by *917the lower court] can be called more compact, the District Court’s preferences do not override whatever state goals were embodied in [the unconstitutional plan]”).
Application of these principles to the situation presented by this stay application is relatively straightforward. Given the status of New Jersey’s redistricting after the 1980 census, the last formal declaration of the State’s policy preferences in congressional reapportionment is contained in the Feldman Plan that we declared unconstitutional last Term. Once the constitutional infirmity in that plan — the unjustified numerical inequality between congressional districts — is remedied, the District Court must choose the alternative plan that remains most faithful to those state policies. We have never concluded, nor in my view should we conclude, that the existence of noncompact or gerrymandered districts is by itself a constitutional violation.2 Cf. Gaffney v. Cummings, 412 U. S. 735, 752-754, and n. 18 (1973). Therefore, absent unconstitutional population variances, or other findings of unconstitutionality such as discrimination on racial or religious lines, the District Court should implement the alternative plan that is most faithful to the districts included in the most recent plan enacted by the State.3
*918In this case, it is clear that the Senate Plan corrects the constitutional violation we found last Term by reducing the numerical inequality to a maximum deviation between districts of 67 people. Cf. Karcher v. Daggett, 462 U. S., at 739-740, n. 10; Simon v. Davis, 463 U. S. 1219 (1983), summarily aff’g In re Pennsylvania Congressional Districts Reapportionment Cases, 567 F. Supp. 1507 (MD Pa. 1982). Moreover, the District Court found, and all parties agree, that the geographical boundaries included in the Senate Plan are closer than those of any alternative plan to the boundaries contained in the unconstitutional Feldman Plan. See 580 F. Supp., at 1262 (the Senate Plan had districts that were “virtually identical” to the districts included in the Feldman Plan); id., at 1264 (districts in another plan were “considerably more compact than those in the Feldman Plan, and thus also more compact than those in [the] Senate [Plan]”). See also Letter from State Attorney General to District Court, p. 1 (Mar. 9, 1984) (“It is quite true . . . that all the parties agree that [the Senate Plan is closer to the Feldman Plan] than any of the proposed alternatives”). Indeed, the Senate Plan would require less than 10% of New Jersey’s residents to change their congressional districts from the 1982 election conducted under the Feldman Plan, whereas the Forsythe Plan adopted by the District Court will require that 31.7% of the State’s residents change districts. Under these circumstances, I believe the District Court erred when it adopted the Forsythe Plan under the mistaken belief that it “owe[d] no deference to an unconstitutional state statute.” 580 F. Supp., at 1263.
Accordingly, I would grant the application for a stay. Moreover, given the imminence of New Jersey’s primary elections, I would remand the case to the three-judge District Court for implementation of the Senate Plan, absent any finding that the Senate Plan, on its own terms, is unconstitutional.
Also before the court were two plans submitted by the Taxpayers Political Action Committee and two plans submitted by representatives of the State’s executive branch. The former plans were considered only briefly by the District Court, and were rejected because their district population variances were “larger than any which would occur in the plans proposed by other parties.” See Daggett v. Kimmelman, 580 F. Supp. 1259, 1262 (NJ 1984). The two plans introduced by the executive branch were considered more extensively, but they too were ultimately rejected. See id., at 1263-1264. None of these plans is currently being pressed for consideration by any party before the Court.
Thus, the factors we noted in Karcher v. Daggett, 462 U. S., at 740 (“making districts compact, respecting municipal boundaries, preserving the cores of prior districts, and avoiding contests between incumbent Representatives”), should be considered only to the extent that a State might rely on them to justify population variances between districts. Indeed, nothing in our opinion last Term was intended to suggest that these factors would necessarily be relevant to the constitutionality of a State’s congressional reapportionment plan in the absence of numerical inequalities that themselves violate the Constitution. If only for this reason, the District Court committed legal error when it adopted these factors as its selection criteria for choosing among alternative remedial plans. Cf. supra, at 914-915.
Hence the District Court also erred when it tried to distinguish this case from White v. Weiser, 412 U. S. 783 (1973), on the ground that “[t]he policy dispute in White v. Weiser among the competing plans was over the district court’s rejection of a state policy of avoiding contests among incumbents,” whereas in this case the Feldman Plan was “designed to produce contests among certain Republican incumbents.” 580 F. Supp., at 1263. As I have noted, n. 2, supra, our finding last Term that the Feldman Plan violated Art. I, §2, of the Constitution was premised on the population disparities found among its districts, and was not intended to suggest that we would reject any of the partisan advantages that may have resulted if the plan were implemented.