dissenting.
The District Court, applying proper legal standards, found that the city of Richmond had failed to prove that its annexation of portions of Chesterfield County, Va., on January 1, 1970, had neither the purpose nor the effect of abridging or diluting the voting rights of Richmond’s black citizens. I believe that that finding, far from being clearly erroneous, was amply supported by the record below, and that the District Court properly denied the declaratory judgment sought by Richmond. I therefore dissent.
I
The Voting Rights Act of 19651 grew out of a long and sorry history of resistance to the Fifteenth Amendment’s ringing proscription of racial discrimination in voting. That history, which we reviewed in the course *380of upholding the Act’s constitutionality in South Carolina v. Katzenbach, 383 U. S. 301, 308-315 (1966), showed a persistent and often ingenious use of tests and devices to disenfranchise black citizens.2 Congress, in response, banned or restricted the use of many of the more familiar discriminatory devices;3 but in addition, recognizing “that some of the States covered by § 4 (b) of the Act had resorted to the extraordinary stratagem of contriving new rules of various kinds for the sole purpose of perpetuating voting discrimination . . . [and] that these States might try similar maneuvers in the future in order to evade the remedies for voting discrimination contained in the Act itself,” 4 Congress enacted the broad prophylactic rule of § 5 of the Act, prohibiting covered States from implementing any new “voting qualification or prerequisite to voting, or standard, practice, or procedure with respect to voting” without first securing the approval of either the Attorney General or the United States District Court for the District of Columbia. In an effort to avoid the delays and uncertainties fostered by prior statutes, under which affected parties or the Attorney General had been forced to assume the initiative in challenging discriminatory voting practices, Congress placed the burden of proof in a § 5 proceeding squarely upon the acting State or municipality to show that its proposed change is free of a racially discriminatory purpose or effect.5 This burden is intended *381to be a substantial one for a State or locality with a history of past racial discrimination.6
In short, Congress, through the Voting Rights Act of 1965, imposed a stringent and comprehensive set of controls upon States falling within the Act’s coverage. We have heretofore held that the language of § 5 was designed “to give the Act the broadest possible scope,” and to require “that all changes, no matter how small, be subjected to § 5 scrutiny,” Allen v. State Board of Elections, 393 U. S. 544, 567-568 (1969); we have thus applied § 5 to legislative reapportionments, annexations, and any other state actions which may potentially abridge or dilute voting rights. Id., at 569-571; Georgia v. United States, 411 U. S. 526 (1973); Perkins v. Matthews, 400 U. S. 379 (1971).
The frontline judicial responsibility for interpreting and applying the substantive standards of § 5 rests exclusively with the United States District Court for the District of Columbia,7 and the considerable experience which that court has acquired in dealing with § 5 cases enhances the respect to which its judgments are entitled on appellate review by virtue of that unique position. The District Court here recognized that it bears a “heavy responsibility” under § 5, and that that “responsibility is no less than to ensure realization of the Fifteenth Amendment’s promise of equal participation in *382our electoral process.” 376 F. Supp. 1344, 1346-1347 (1974). In exercising our power of appellate review over that court’s substantive § 6 determinations, we must be equally devoted to that same majestic promise.
II
In my view, the flagrantly discriminatory purpose with which Richmond hastily settled its Chesterfield County annexation suit in 1969 compelled the District Court to deny Richmond the declaratory judgment. The record is replete with statements by Richmond officials which prove beyond question that the predominant (if not the sole) motive and desire of the negotiators of the 1969 settlement was to acquire 44,000 additional white citizens for Richmond, in order to avert a transfer of political control to what was fast becoming a black-population majority.8 The District Court’s findings on this point were quite explicit:
“Richmond’s focus in the negotiations was upon the number of new white voters it could obtain by annexation; it expressed no interest in economic or geographic considerations such as tax revenues, vacant land, utilities, or schools. The mayor required assurances from Chesterfield County officials that at least 44,000 additional white citizens would be obtained by the City before he would agree upon settlement of the annexation suit. And the mayor and one of the city councilmen conditioned final acceptance of the settlement agreement on the annexation going into effect in sufficient time to make citizens in the annexed area eligible to vote in the City Council elections of 1970.”9
*383Against this background, the settlement represented a clear victory for Richmond’s entrenched white political establishment: the city realized a net gain of 44,000 white citizens, its black population was reduced from 52% to 42% of the total population, and the predominantly white Richmond Forward organization retained its 6-3 majority on the city council.
Having succeeded in this patently discriminatory enterprise, Richmond now argues that it can purge the taint of its impermissible purpose by dredging up supposed objective justifications for the annexation and by replacing its practice of at-large councilmanic elections with a ward-voting system. The implications of the proposed ward-voting system are discussed in Part III, infra; meanwhile, I have grave difficulty with the idea that the taint of an illegal purpose can, under § 5, be dispelled by the sort of post hoc rationalization which the city now offers.
The court below noted that Richmond, in initiating annexation proceedings in 1962, was motivated “by legitimate goals of urban expansion.” 376 F. Supp., at 1351. By 1969, however, those legitimate goals had been pushed into the background by the unseemly haste of the white political establishment to protect and solidify its position of power. The District Court’s findings quoted above fully establish that the 1969 settlement of Richmond’s annexation suit was negotiated in an atmosphere totally devoid of any concern for economic or administrative issues; the city’s own Boundary Expansion Coordinator was not even consulted about the financial or geographical implications of the so-called Horner-Bagley line until several weeks after the line had been drawn.10 The contours of this particular annexation were shaped solely by racial and political considerations, *384and the inference is not merely reasonable but indeed compelled that the annexation line would have been significantly different had the racial motivation not been present.11
To hold that an annexation agreement reached under such circumstances can be validated by objective economic justifications offered many years after the fact, in my view, wholly negates the prophylactic purpose of § 5.12 The Court nevertheless, at the suggestion of the United States, remands for the taking of further evidence on the presence of any “objectively verifiable, legitimate reasons for the annexation.” Even assuming, as the District Court did, that such reasons could now validate an originally illegal annexation, I cannot agree that a remand is necessary.
The District Court, adopting the findings of the Master whom it had appointed under Fed. Rule Civ. Proc. 53, squarely held that Richmond “ 'has failed to establish any counterbalancing economic or administrative benefits of the annexation.’ ” 376 F. Supp., at 1353. The *385record before the Master, including the entire record in Holt v. City of Richmond, 334 F. Supp. 228 (ED Va. 1971), rev’d, 459 F. 2d 1093 (CA4), cert. denied, 408 U. S. 931 (1972), to which the parties stipulated,13 contained ample evidence on the economic and administrative consequences of the annexation. The Master and the District Court weighed this often conflicting evidence and found that Richmond had failed to carry its burden of proof by showing any legitimate purpose for the annexation as consummated in 1969.14
Federal Rule Civ. Proc. 52 (a) compels us to accept that finding unless it can be called clearly erroneous. I find it impossible, on this record, to attach that label to the findings below, and indeed, the Court never goes so far as to do so. Nevertheless, in apparent disagreement with the manner in which conflicting evidence was weighed and resolved by the lower court, the Court remands for further evidentiary proceedings, perhaps in hopes that a re-evaluation of the evidence will produce a more acceptable result. This course of action is to me wholly inconsistent with the proper role of an appellate court operating under the strictures of Rule 52 (a).
Ill
The second prong of any § 5 inquiry is whether the voting change under consideration will have the effect of denying or abridging the right to vote on account of *386race or color. In Perkins v. Matthews, supra, holding that § 5 applies to annexations, we said:
“Clearly, revision of boundary lines has an effect on voting in two ways: (1) by including certain voters within the city and leaving others outside, it determines who may vote in the municipal election and who may not; (2) it dilutes the weight of the votes of the voters to whom the franchise was limited before the annexation, and ‘the right of suffrage can be denied by a debasement or dilution of the weight of a citizen’s vote just a,s effectively as by wholly prohibiting the free exercise of the franchise.’ Reynolds v. Sims, 377 U. S. 533, 555 (1964). Moreover, § 5 was designed to cover changes having a potential for racial discrimination in voting, and such potential inheres in a change in the composition of the electorate affected by an annexation.” 400 U. S., at 388-389.
The guidelines of this discussion in Perkins were correctly applied by the District Court, which continued as follows:
“Perkins left implicit the obvious: If the proportion of blacks in the new citizenry from the annexed area is appreciably less than the proportion of blacks living within the city’s old boundaries, and particularly if there is a history of racial bloc voting in the city, the voting power of black citizens as a class is diluted and thus abridged.” 376 F. Supp., at 1348 (footnote omitted).
Measured against these standards, the dilutive effect of Richmond’s annexation is clear, both as a matter of semantics and as a matter of political realities. Blacks constituted 52% of the preannexation population and 44.8% of the preannexation voting-age population in *387Richmond, but now constitute only 42%. of the postannexation population and only 37.3% of the postannexation voting-age population. I cannot agree that such a significant dilution of black voting strength can be remedied, for § 5 purposes, simply by allocating to blacks a reasonably proportionate share of voting power within the postannexation community.
The history of the Voting Rights Act, as set forth in Part I, supra, discloses the intent of Congress to impose a stringent system of controls upon changes in state voting practices in order to thwart even the most subtle attempts to dilute black voting rights. We have elsewhere described the Act as “an unusual, and in some aspects a severe, procedure for insuring that States would not discriminate on the basis of race in the enforcement of their voting laws.” 15 Congress was certainly aware of the hardships and inconvenience which § 5 and other portions of the Act could impose upon covered States, and localities; but in passing the Act in its final form, Congress unmistakably declared that those hardships are outweighed by the need to ensure effective protection for black voting rights.
Today’s decision seriously weakens the protection so emphatically accorded by the Act. Municipal politicians who are fearful of losing their political control to emerging black voting majorities are today placed on notice that their control can be made secure as long as they can find concentrations of white citizens into which to expand their municipal boundaries. Richmond’s black population, having finally begun to approach an opportunity to elect responsive officials and to have a significant voice in the conduct of its municipal affairs, now finds its voting strength reduced by a plan which “guarantees” *388four seats on the City Council but which makes the elusive fifth seat more remote than it was before. The Court would offer, as consolation, the fact that blacks will enjoy a fair share of the voting power available under a ward system operating within the boundaries of the postannexation community; but that same rationale would support a plan which added far greater concentrations of whites to the city and reduced black voting strength to the equivalent of three seats, two seats, or even fractions of a seat. The reliance upon postannexation fairness of representation is inconsistent with what I take to be the fundamental objective of § 5, namely, the protection of present levels of voting effectiveness for the black population.
It may be true, as the Court suggests, that this interpretation would effectively preclude some cities from undertaking desperately needed programs of expansion and annexation. Certainly there is nothing in § 5 which suggests that black voters could or should be given a disproportionately high share of the voting power in a postannexation community; where the racial composition of an annexed area is substantially different from that of the annexing area, it may well be impossible to protect preannexation black voting strength without invidiously diluting the voting strength of other racial groups in the community. I see no reason to assume that the “demographics” of the situation are such that this would be an insuperable problem for all or even most cities covered by the Act; but in any event, if there is to be a “municipal hardship” exception for annexations vis-a-vis § 5, that exception should originate with Congress and not with the courts.
At the very least, therefore, I would adopt the Peters-burg standard relied upon by the District Court, namely, that the dilutive effect of an annexation of this sort can *389be cured only by a ward plan “ 'calculated to neutralize to the extent possible any adverse effect upon the political participation of black voters.' ” 376 F. Supp., at 1352.16 The Crusade for Voters of Richmond, int'ervenor in the court below, submitted several plans providing for a greater black representation in the so-called “swing district” than that afforded by Richmond's own plan; the District Court, in light of these alternative submissions and in light of the fact that Richmond's ward plan had been drawn up without any reference to racial living patterns, concluded that Richmond’s plan did not, “to the extent possible,” minimize dilution of black voting power. Id., at 1356-1357. On that basis, I would affirm the finding that Richmond failed to establish the absence of a discriminatory effect prohibited by § 5.
IV
More than five years have elapsed since the last municipal elections were held in Richmond.17 Hopes which were lifted by the District Court decision over a year ago are today again dashed, as the case is remanded for what may prove to be several additional years of litigation; Richmond will continue to be governed, as it has been for the last five years, by a slate of councilmen elected in clear violation of § 5.18 The black population of Richmond may be justifiably suspicious of the “pro*390tection” its voting rights are receiving when these rights can be suspended in limbo, and the people deprived of the right to select their local officials in an election meeting constitutional and statutory standards, for so many years. I would affirm the judgment below, and let the United States District Court for the Eastern District of Virginia set about the business of fashioning an appropriate remedy as expeditiously as possible.
79 Stat. .437, as amended, 84 Stat. 314, 42 U. S. C. § 1973 et seq.
See also Beer v. United States, 374 F. Supp. 363, 377-378 (DC 1974); H. R. Rep. No. 439, 89th Cong., 1st Sess., 8-13 (1965); S. Rep: No. 162, pt. 3, 89th Cong., 1st Sess., 3-12 (1965).
These devices included literacy tests, requirements of “good moral character,” and voucher requirements, §§ 4 (a)-(d), 42 U. S. C. §§ 1973b (a)-(d), as well as poll taxes, § 10, 42 U. S. C. § 1973h.
South Carolina v. Katzenbach, 383 U. S. 301, 335 (1966).
Georgia v. United States, 411 U. S. 526, 538 (1973).
City of Petersburg v. United States, 354 F. Supp. 1021, 1027 (DC 1972), aff’d, 410 U. S. 962 (1973).
We have consistently held that the substantive issue of discriminatory purpose or effect under § 5 can be litigated only in the District Court for the District of Columbia; the sole question open for consideration in any other district court is whether a state voting practice or requirement is of the sort required by § 5 to be submitted for prior approval. Perkins v. Matthews, 400 U. S. 379, 383-386 (1971); Allen v. State Board of Elections, 393 U. S. 544, 555-559 (1969); Connor v. Waller, 421 U. S. 656 (1975).
376 F. Supp. 1344, 1349-1350 (DC 1974), The statements quoted, id., at 1349 n. 29, particularly those of then-Mayor Bagley, can hardly be described as subtle or indirect.
Id., at 1350 (footnotes omitted).
2 App. 352-354.
Several judges involved in a prior phase of this dispute have expressed a belief, founded upon the record, that Richmond would have secured far more favorable annexation terms had it not been prodded into a hasty settlement by the pendency of the 1970 elections. See Holt v. City of Richmond, 459 F. 2d 1093, 1108 (CA4) (Winter, J., dissenting), cert. denied, 408 U. S. 931 (1972); Holt v. City of Richmond, 334 F. Supp. 228, 236 (ED Va. 1971), rev’d on other grounds, 459 F. 2d 1093, supra.
Had this agreement been properly submitted for § 5 clearance in 1969, I cannot believe that the annexation would ever have been permitted to take place. But our holding in Perkins v. Matthews, supra, that annexations fall within the scope of § 5, came more than a year after the Richmond annexation took effect; by this quirk of timing, the annexation escaped preimplementation scrutiny entirely. The 1969 line thus remains in place, a grim reminder in its contours and in its very existence of the discriminatory purpose which gave it birth.
376 F. Supp., at 1349.
Much of the evidence in the record below appears to have dealt with Richmond’s need for expansion and annexation in the abstract. Annexation in the abstract, however, is not at issue here; the critical question is whether the particular line drawn in 1969 had any contemporary justification in terms of objective factors such as Richmond’s need for vacant land, an expanded tax base, and the like.
Allen v. State Board of Elections, 393 U. S., at 556 (footnote omitted).
The original version of this standard appears in City of Peters-burg v. United States, 354 F. Supp., at 1031.
The last councilmanie election was held on June 10,1970. 1 App. 71; 376 F. Supp., at 1351.
The 1970 elections were conducted on an at-large basis in the postannexation community, a procedure inconsistent with even the narrowed Petersburg “effect” test adopted by the Court today. Moreover, since the elections occurred prior to our decision in Perkins, supra, there was no attempt to submit the annexation for prior approval. Section 5 is violated in both respects.