JUSTICE SOUTER, with whom JUSTICE STEVENS, JUSTICE GINSBURG, and JUSTICE BREYER join, concurring in part and dissenting in part.

Under § 5 of the Voting Rights Act of 1965, 42 U.S.C. § 1973c, a jurisdiction required to obtain preclearance of changes to its voting laws must show that a proposed amendment will not have the effect, and does not reflect a purpose, to deny or abridge the vote on account of race. I respectfully dissent from the Court's holding that § 5 is indifferent to a racially discriminatory purpose so long as a change in voting law is not meant to diminish minority voting strength below its existing level. It is true that today's decision has a precursor of sorts in Beer v. United States, 425 U.S. 130 (1976), which holds that the only anticipated redistricting effect sufficient to bar preclearance is retrogression in minority voting strength, however dilutive of minority voting power a redistricting plan may otherwise be. But if today's decision achieves a symmetry with Beer, the achievement is merely one of well-matched error. The Court was mistaken in Beer when it restricted the effect prong of § 5 to retrogression, and the Court is even more wrong today when it limits the clear text of § 5 to the corresponding retrogressive purpose. Although I adhere to the strong policy of respecting precedent in statutory interpretation and so would not reexamine Beer, that policy does not demand that recognized error be compounded indefinitely, and the Court's prior mistake about the meaning of the effects requirement of § 5 should not be expanded by an even more erroneous interpretation of the scope of the section's purpose prong.

The Court's determination that Congress intended preclearance of a plan not shown to be free of dilutive intent (let alone a plan shown to be intentionally discriminatory) is not, however, merely erroneous. It is also highly unconvincing. The evidence in these very cases shows that the Bossier Parish School Board (School Board or Board) acted with intent to dilute the black vote, just as it acted with that same intent through decades of resistance to a judicial desegregation order. The record illustrates exactly the sort of relentless bad faith on the part of majority-white voters in covered jurisdictions that led to the enactment of § 5. The evidence all but poses the question why Congress would ever have meant to permit preclearance of such a plan, and it all but invites the answer that Congress could hardly have intended any such thing. While the evidence goes substantially unnoticed on the Court's narrow reading of the purpose prong of § 5, it is not only crucial to my resolution of these cases, but insistent in the way it points up the implausibility of the Court's reading of purpose under § 5.

I

In Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252 (1977), this Court set out a checklist of considerations for assessing evidence going to discriminatory intent: the historical background of a challenged decision, its relative impact on minorities, specific antecedent events, departures from normal procedures, and contemporary statements of decisionmakers. Id., at 266-268. We directed the District Court to follow that checklist in enquiring into discriminatory intent following remand in these cases, Reno v. Bossier Parish School Bd., 520 U.S. 471, 488 (1997) (Bossier Parish I). The Arlington Heights enquiry reveals the following account of the School Board's redistricting activity and of the character of the parish in which it occurred.

The parish's institution of general governance is known as the Police Jury, a board of representatives chosen from districts within the parish. After the 1990 census showed a numerical malapportionment among those districts, the Police Jury prepared a revised districting plan, which they submitted to the Attorney General of the United States with a request for the preclearance necessary under § 5 of the Voting Rights Act before the parish, a covered jurisdiction, could modify its voting district lines. Based on information then available to the Department of Justice, the Attorney General understood the parish to have shown that the new plan would not have the effect and did not have the purpose of abridging the voting rights of the parish's 20% black population, and the revised Police Jury plan received preclearance in the summer of 1991. In fact, as the parish's School Board has now admitted, the Police Jury plan thus approved dilutes the voting strength of the minority population, Plaintiff's Brief on Remand 12; that is, the plan discriminates by abridging the rights of minority voters to participate in the political process and elect candidates of their choice. Thornburg v. Gingles, 478 U.S. 30, 46-47 (1986).

The same population shifts that required the Police Jury to reapportion required the elected School Board to do the same. Although the Board had approached the Police Jury about the possibility of devising a joint plan of districts common to both Board and jury, the jury rebuffed the Board, see App. to Juris. Statement 172a (Stipulations 83-84), and the Board was forced to go it alone. History provides a good indication of what might have been expected from this endeavor.

As the parties have stipulated, the School Board had applied its energies for decades in an effort to "limit or evade" its obligation to desegregate the parish schools. Id., at 216a (Stipulation 237). When the Board first received a court order to desegregate the parish's schools in the mid-1960's, it responded with the flagrantly defiant tactics of that era, see id., at 216a-217a (Stipulations 236-237), and the record discloses the Board's continuing obstructiveness down to the time covered by these cases. During the 1980's, the degree of racial polarization in the makeup of the parish's schools rose, id., at 218a (Stipulations 241-243), and the disproportionate assignment of black faculty to predominantly black schools increased, id., at 217a-218a (Stipulation 240). While the parish's superintendent testified that the assignment of black faculty to predominantly black schools came in response to black parents' requests for positive black examples for their children, see App. 289, the black leaders who testified in these cases uniformly rejected that claim and insisted that, in accord with the parish's desegregation decree, black faculty were to be distributed throughout the parish's schools, to serve as models for white, as well as black, students, see id., at 326-327; 2 Tr. 126-128.

Other evidence of the Board's intransigence on race centers on the particular terms of the integration decree that since 1970 has required the Board to maintain a "Bi-Racial Advisory Review Committee" made up of an equal number of black and white members in order to "'recommend to the . . . Board ways to attain and maintain a unitary system and to improve education in the parish.'" App. to Juris. Statement in No. 98-405, p. 182a (Stipulation 111) (hereinafter App. to Juris. Statement). Although the Board represented to the District Court overseeing desegregation that the committee was in place, see 2 Tr. 16 (testimony of Superintendent William T. Lewis), the committee actually met only two or three times in the mid-1970's and then with only its black members in attendance, see App. to Juris. Statement 183a (Stipulation 112). In 1993, the Board set up a short-lived "Community Affairs Committee" to replace the "Bi-Racial Committee." Despite the Board's resolution charging the committee "'with the responsibility of investigating, consulting and advising the court and school board periodically with respect to all matters pertinent to the retention [sic] of a unitary school system,'" ibid. (Stipulation 114), the Board disbanded the committee after only three months because, as a leading Board member put it, "'the tone of the committee made up of the minority members of the committee quickly turned toward becoming involved in policy,'" id., at 184a (Stipulation 116). "Policy," however, was inevitably implicated by the committee's purpose, and the subjects of its recommendations (such as methods for more effective recruitment of black teachers and their placement throughout the school system in accord with the terms of the desegregation decree, see id., at 183a-184a (Stipulation 115)) fell squarely within its mandate. It is thus unsurprising that the Board has not achieved a unitary school system and remains under court order to this day. See id., at 217a (Stipulation 239); App. 139 (testimony of S. P. Davis).

About the time the Board appointed its "Community Affairs Committee," it sought preclearance under § 5 from the Attorney General for the redistricting plan before us now. The course of the Board's redistricting efforts tell us much about what it had in mind when it proposed its plan. Following the rebuff from the Police Jury, the Board was able to follow a relaxed redistricting timetable, there being no Board elections scheduled before 1994. While the Board could simply have adopted the Police Jury plan once the Attorney General had precleared it, the Board did not do so, App. to Juris. Statement 147a (Stipulation 11), despite just such a proposal from one Board member at the Board's September 5, 1991, meeting. No action was then taken on the proposal, id., at 174a (Stipulations 89-90), and although the Board issued no explanation for its inaction, it is noteworthy that the jury plan ignored some of the Board's customary districting concerns. Whereas one of those concerns was incumbency protection, see App. 251; cf. App. to Juris. Statement 152a (Stipulation 26), the jury plan would have pitted two pairs of incumbents against each other and created two districts in which no incumbent resided. id., at 181a-182a (Stipulation 109). The jury plan disregarded school attendance zones, and even included two districts containing no schools. Id., at 174a, 151a, 191a (Stipulations 88, 24, 141). The jury plan, moreover, called for a total variation in district populations exceeding the standard normally used to gauge satisfaction of the "one person, one vote" principle, see id., at 162a-163a (Stipulation 58); App. 231-232; 1 Tr. 147, four of its districts failed the standard measure of compactness used by the Board's own cartographer, id., at 174-176, and one of its districts contained noncontiguous elements, App. 234-235.

In addressing the need to devise a plan of its own, the Board hired the same redistricting consultant who had advised the Police Jury, Gary Joiner. Joiner and the Board members (according to Joiner's testimony) were perfectly aware of their responsibility to avoid vote dilution in accordance with the Voting Rights Act, see Record, Doc. No. 38 (direct testimony of Joiner 5), and he estimated that it would take him between 200 to 250 hours to devise a plan for the Board. The Board then spent nearly a year doing little in public about redistricting, while its members met in private with Joiner to consider alternatives. In March 1992, George Price, president of the parish's branch of the National Association for the Advancement of Colored People (NAACP), wrote to the superintendent of parish schools asking for a chance to play some role in the redistricting process. App. 184. Although the superintendent passed the letter on to the Board, the Board took no action, and neither the superintendent nor the Board even responded to Price's request. App. to Juris. Statement 175a (Stipulation 93). In August, Price wrote again, this time in concert with a number of leaders of black community organizations, again seeking an opportunity to express views about the redistricting process, as well as about a number of Board policies bearing on school desegregation. App. 187-189; see also App. to Juris. Statement 175a (Stipulation 94). Once again the Board made no response.

Being frustrated by the Board's lack of responsiveness, Price then asked for help from the national NAACP's Redistricting Project, which sent him a map showing how two compact majority-black districts might be drawn in the parish. Id., at 177a (Stipulation 98). When Price showed the map to a school district official, he was told it was unacceptable because it failed to show all 12 districts. At Price's request, the Redistricting Project then provided a plan showing all 12 districts, which Price presented to the Board at its September 3, 1992, meeting, explaining that it showed the possibility of drawing majority-black districts. Id., at 177a-178a (Stipulations 99-100). Several Board members said they could not consider the NAACP plan unless it was presented on a larger map, id., at 178a (Stipulation 100), and both the Board's cartographer and its legal advisor, the parish district attorney, dismissed the plan out of hand because it required precinct splits, id., at 179a (Stipulation 102).

There is evidence that other implications of the NAACP proposal were objectionable to the Board. According to one black leader, Board member Henry Burns told him that while he personally favored black representation on the Board, a number of other Board members opposed the idea. App. 142. According to George Price, Board member Barry Musgrove told him that the Board was hostile to the creation of a majority-black district. Id., at 182.

Although the NAACP plan received no further public consideration, the pace of public redistricting activity suddenly speeded up. At the Board's September 17, 1992, meeting, without asking Joiner to address the possibility of creating any majority-black district, the Board abruptly passed a statement of intent to adopt the Police Jury plan. App. to Juris. Statement 179a-180a (Stipulation 106). At a public hearing on the plan one week later, attended by an overflow crowd, a number of black voters spoke against the plan, and Price presented the Board with a petition bearing over 500 signatures urging consideration of minority concerns. No one spoke in favor of the plan, Bossier Parish I, 907 F. Supp. 434, 439 (DC 1995), and Price explained to the Board that preclearance of the jury plan for use by the Police Jury was no guarantee of preclearance of the same plan for the Board. App. to Juris. Statement 180a-181a (Stipulation 108). Nonetheless, at its October 1 meeting, the voting members of the Board unanimously adopted the Police Jury plan, with one member absent and the Board's only black member (who had been appointed just two weeks earlier to fill a vacancy) abstaining. Id., at 181a-182a (Stipulation 109). The Board did not submit the plan for preclearance by the Attorney General until January 4, 1993. Id., at 182a (Stipulation 110).

II

The significance of the record under § 5 is enhanced by examining in more detail several matters already mentioned as free from dispute, by testing some of the Board's stated reasons for refusing to consider any NAACP plan, and by looking critically at the District Court's reasons for resolving disputed issues in the School Board's favor.

A

The parties stipulate that for decades before this redistricting the Board had sought to "limit or evade" its obligation to end segregation in its schools, an obligation specifically imposed by Court order nearly 35 years ago and not yet fulfilled. The Board has also conceded the discriminatory impact of the Police Jury plan in falling "more heavily on blacks than on whites," Plaintiff's Brief on Remand in Civ. Action No. 94-1495 (D.D.C.), p. 12, and in diluting "black voting strength," id., at 21. Even without the stipulated history, the conceded dilution would be evidence of a correspondingly discriminatory intent. With the history, the implication of intent speaks louder, and it grows more forceful still after a closer look at two aspects of the dilutive impact of the Police Jury plan.

First, the plan includes no majority-black districts even though residential and voting patterns in Bossier Parish meet the three conditions we identified in Thornburg v. Gingles, 478 U.S., at 50-51, as opening the door to drawing majority-minority districts to put minority voters on an equal footing with others. The first Gingles condition is that "the minority group must be able to demonstrate that it is sufficiently large and geographically compact to constitute a majority in a single-member district." Id., at 50. The Board does not dispute that black voters in Bossier Parish satisfy this criterion. The Board joined in a stipulation of the parties that in 1991, "it was obvious that a reasonably compact majority-black district could be drawn within Bossier City," App. to Juris. Statement 154a-155a (Stipulation 36); see also 1 Tr. 60 (statement of Board member Barry Musgrove), and that the NAACP plan demonstrated that two such districts could have been drawn in the parish, see App. to Juris. Statement 192a (Stipulation 143). As to the second and third Gingles conditions, that the minority population be politically cohesive and that the majority-white block voting be enough to defeat the minority's preferred candidate, see Gingles, supra, at 51, the Government introduced expert testimony showing such polarization in Bossier Parish's voting patterns. See App. to Juris. Statement 201a- 207a (Stipulations 181-196); App. 163-173 (declaration of Dr. Richard Engstrom). While acknowledging the somewhat limited data available for analysis, the expert concluded that "African American voters are likely to have a realistic opportunity to elect candidates of their choice to the . . . Board only in districts in which they constitute a majority of the voting age population." Id., at 174.

Second, the Police Jury plan diluted black votes by dividing neighboring black communities with common interests in and around at least two of the Parish's municipalities, thereby avoiding the creation of a majority-black district. See id., at 154-156 (declaration of George J. Castille III); id., at 141 (testimony of S. P. Davis). Even the Board's own cartographer conceded that one of these instances "'appear[ed]'" to constitute "'fracturing,'" App. to Juris. Statement 191a (Stipulation 138), which he defined as "divid[ing] a 'population that has a traditional cohesiveness, lives in the same general area, [and] has a lot of commonalties' . . . with '[the] intent to . . . fracture that population into adjoining white districts,'" id., at 189a-190a (Stipulation 133).

B

The Board's cartographer and lawyer objected that the NAACP plan was unacceptable because it split precincts in violation of state law. And yet the Board concedes that school boards were free to seek precinct changes from the police juries of their parishes, as they often successfully did. See id., at 150a-151a (Stipulations 22-23). One of the Government's experts, see App. 214, 217, 354, and the Board's own cartographic consultant, see App. to Juris. Statement 151a (Stipulation 23), acknowledged this practice. Indeed, the parties agree that Joiner advised the Board about the option of going to the Police Jury for precinct changes, see id., at 174a (Stipulation 89); see also id., at 179a (Stipulation 102), but that the Board never asked him to pursue that possibility, see id., at 188a (Stipulation 128). Judge Kessler in the District Court was therefore surely correct that the Board's claimed inability to divide precincts was no genuine obstacle to a plan with a majority-black district. See Bossier Parish I, 907 F. Supp., at 460-461 (opinion concurring in part and dissenting in part).

It becomes all the clearer that the prospect of splitting precincts was no genuine reason to reject the NAACP plan (or otherwise to refuse to consider creating any majority-black districts) when one realizes that from early on in the Board's redistricting process it gave serious thought to adopting a plan that would have required just such precinct splits. When the Board hired Joiner as its cartographer in May 1991, his estimate of 200 to 250 hours to prepare a plan for the Board, see App. to Juris. Statement 173a (Stipulation 86), indicated that there was no intent simply to borrow the recently devised Police Jury plan or to build on the precincts established by the Police Jury, a possibility that Joiner thought could be explored in "[s]everal hours at least," App. 271. It seems obvious that from the start the Board expected its plan to require precinct splitting, and Joiner acknowledged in his testimony that any plan "as strong as" the Police Jury plan in terms of traditional districting criteria would require precinct splits. Ibid. Splitting precincts only became an insuperable obstacle once the NAACP made its proposal to create majority-black districts.

C 1

Despite its stated view that the record would not support a conclusion of nonretrogressive discriminatory intent, the District Court majority listed a series of "allegedly dilutive impacts" said to point to discriminatory intent: "[t]hat some of the new districts have no schools, that the plan ignores attendance boundaries, that it does not respect communities of interest, that there is one outlandishly large district, that several of them are not compact, that there is a lack of contiguity, and that the population deviations resulting from the jury plan are greater than the limits (± 5%) imposed by Louisiana law." 7 F. Supp.2d 29, 32 (DC 1998) (Bossier Parish II). The District Court found this evidence "too theoretical, and too attenuated," to be probative of retrogressive intent in the absence of corroborating evidence of a "deliberate attempt." Ibid. But whatever the force of such evidence may be on the issue of intent to cause retrogression, there is nothing "theoretical" or "attenuated" in its significance as showing intent to dilute generally.

2

If we take the District Court opinions in Bossier Parish I and Bossier Parish II together and treat the court's § 5 discussions as covering nonretrogressive discriminatory intent, it is clear that the court rested on two reasons for finding that the plan's dilutive effect could not support an inference of nonretrogressive discriminatory intent. First, the court thought any such inference inconsistent with the view expressed in Miller v. Johnson, 515 U.S. 900, 924 (1995), that a refusal to adopt a plan to maximize the number of majority-minority districts is insufficient alone to support an inference of intentional discrimination. Miller is not on point, however. In Miller, Georgia had already adopted a plan that clearly improved the position of minority voters by establishing two majority-black districts. The question was simply whether the State's refusal to create a third betrayed discriminatory intent. Id., at 906-908, 923-924. In these cases, the issue of inferred intent did not arise upon rejection of a plan maximizing the number of majority-black districts after a concededly ameliorative plan had already been adopted; the issue arose on the Board's refusal to consider a plan with any majority-black districts when more than one such district was possible under Gingles. The issue here is not whether Bossier Parish betrayed a discriminatory purpose in refusing to create the maximum number of majority-black districts, see Bossier Parish II, supra, at 33 (Silberman, J., concurring), but simply whether it was significant that the parish refused to consider creating a majority-black district at all. The refusal points to a discriminatory intent that the refusal to maximize in Miller v. Johnson did not show.

The District Court's second ground for discounting the evidence of intent inherent in the Police Jury plan's dilutive effect was its finding that the Board had legitimate, nondiscriminatory reasons for approving the plan. The evidence, however, is powerful in showing that the Board had no such reasons. As I have already noted, the Board's respect for existing precinct lines was apparently pretextual. The other supposedly legitimate reason for the Board's choice, that the Police Jury was a safe harbor under § 5, is equally unlikely. If the Police Jury plan was a safe harbor, it had been safe from the day the Attorney General precleared it for the Police Jury, whereas the Board ignored it for more than a year after that preclearance. Interest in the Police Jury plan developed only after pressure from Price and the NAACP had intensified to the point that the redistricting process would have to be concluded promptly if the minority proposals were not to be considered. The Police Jury, therefore, became an attractive harbor only when it seemed to offer safety from demands for a fair reflection of minority voting strength. It was chosen by a Board, described by the District Court majority as possessing a "tenacious determination to maintain the status quo," Bossier Parish II, supra, at 32, and the only fair inference is that when the Board suddenly embraced the Police Jury plan it was running true to form.

D

In sum, for decades the School Board manifested sedulous resistance to the constitutional obligation to desegregate parish schools, which have never attained unitary status and are still subject to court order. When faced with the need to act alone in redrawing its voting districts, the Board showed no interest in the Police Jury plan, which made no sense for school purposes and was at odds with normal districting principles applied by the Board. The Board hired a cartographer in anticipation of drawing district lines significantly different from the Police Jury lines, and the Attorney General's preclearance of the Police Jury plan for the jury's use produced no apparent Board interest in adopting that same plan. When minority leaders sought a role in proposing a plan, the Board ignored them and when they produced concrete proposals prepared by the NAACP, the Board sidestepped with successive technical reasons culminating in a patently pretextual objection. It was only then, as its pretexts for resisting the NAACP were wearing thin, that the Board evidently scrapped its intention to obtain an original plan tailored to school district concerns and acted with unwonted haste on the year-old proposal to adopt the manifestly unsuitable Police Jury plan. The proposal received no public hearing support and nothing but objection from minority voters, who pointed out what the Board now agrees, that the Police Jury plan dilutes minority voting strength. The objections were unavailing and the Board adopted the dilutive plan.

There is no reasonable doubt on this record that the Board chose the Police Jury plan for no other reason than to squelch requests to adopt the NAACP plan or any other plan reflecting minority voting strength, and it would be incredible to suggest that the resulting submergence of the minority voters was unintended by the Board whose own expert testified that it understood the illegality of dilution. If, as I conclude below, see Part III, infra, dilutive but nonretrogressive intent behind a redistricting plan disqualifies it from § 5 preclearance, then preclearance is impossible on this record. Since the burden to negate such intent (like the burden to negate retrogressive intent and effect) rests on the voting district asking for preclearance, nothing more is required to show the impossibility of preclearance. See, e.g., Pleasant Grove v. United States, 479 U.S. 462, 469 (1987). It is worth noting, however, that the parish should likewise lose even if we assume, as the District Court majority seems to have done at one point, that the burden to show disqualifying intent is on the Government and the intervenors. Bossier Parish II, 7 F. Supp.2d, at 31 ("We can imagine a set of facts that would establish a 'non-retrogressive, but nevertheless discriminatory purpose,' but those imagined facts are not present here"). It is not only that Judge Kessler was correct in her conclusion that dilutive but nonretrogressive intent was shown; the contrary view of the District Court majority raises "'the definite and firm conviction that a mistake [has] been committed,'" Concrete Pipe Products of Cal., Inc. v. Construction Laborers Pension Trust for Southern Cal., 508 U.S. 602, 622 (1993) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395 (1948)). Regardless of the burden of persuasion, therefore, the parish should lose under the intent prong of § 5, if the purpose that disqualifies under § 5 includes an intent to dilute minority voting strength regardless of retrogression.

III A

The legal issue here is the meaning of "abridging" in the provision of § 5 that preclearance of a districting change in a covered jurisdiction requires a showing that the new plan does not "have the purpose . . . of denying or abridging the right to vote on account of race or color . . . ." The language tracks that of the Fifteenth Amendment's guarantee that "[t]he right of citizens . . . to vote shall not be denied or abridged . . . on account of race [or] color . . . ." Since the Act is an exercise of congressional power under § 2 of that Amendment, South Carolina v. Katzenbach, 383 U.S. 301, 325-327 (1966), the choice to follow the Amendment's terminology is most naturally read as carrying the meaning of the constitutional terms into the statute. United States v. Kozminski, 487 U.S. 931, 945 (1988) ("By employing the constitutional language, Congress apparently was focusing on the prohibition of comparable conditions"); cf. Morissette v. United States, 342 U.S. 246, 263 (1952) ("[W]here Congress borrows terms of art in which are accumulated the legal tradition and meaning of centuries of practice, it presumably knows and adopts the cluster of ideas that were attached to each borrowed word in the body of learning from which it was taken and the meaning its use will convey to the judicial mind unless otherwise instructed"). Any construction of the statute, therefore, carries an implication about the meaning of the Amendment, absent some good reason to treat the parallel texts differently on some particular point, and a reading of the statute that would not fit the Constitution is presumptively wrong.

In each context, it is clear that abridgment necessarily means something more subtle and less drastic than the complete denial of the right to cast a ballot, denial being separately forbidden. Abridgment therefore must be a condition in between complete denial, on the one hand, and complete enjoyment of voting power, on the other. The principal concept of diminished voting strength recognized as actionable under our cases is vote dilution, defined as a regime that denies to minority voters the same opportunity to participate in the political process and to elect representatives of their choice that majority voters enjoy. See, e.g., Thornburg v. Gingles, 478 U.S., at 46-47; 42 U.S.C. § 1973. The benchmark of dilution pure and simple is thus a system in which every minority voter has as good a chance at political participation and voting effectiveness as any other voter. Our cases have also recognized retrogression as a subspecies of dilution, the consequence of a scheme that not only gives a minority voter a lesser practical chance to participate and elect than a majority voter enjoys, but even reduces the minority voter's practical power from what a preceding scheme of electoral law provided. See Beer v. United States, 425 U.S., at 141. Although our cases have dealt with vote dilution only under the Fourteenth Amendment, see, e.g., Shaw v. Reno, 509 U.S. 630, 645 (1993), I know of no reason in text or history that dilution is not equally violative of the Fifteenth Amendment guarantee against abridgment. And while there has been serious dispute in the past over the Fourteenth Amendment's coverage of voting rights, see, e.g., Oregon v. Mitchell, 400 U.S. 112, 154 (1970) (Harlan, J., concurring in part and dissenting in part), I know of no reason to doubt that "abridg[e]" in the Fifteenth Amendment includes dilutive discrimination. See Bossier Parish I,

520 U.S., at 494-495 (BREYER, J., concurring in part and concurring in judgment).

The Court has never held (save in Beer) that the concept of voting abridgment covers only retrogressive dilution, and any such reading of the Fifteenth Amendment would be outlandish. The Amendment contains no textual limitation on abridgment, and when it was adopted, the newly emancipated citizens would have obtained practically nothing from a mere guarantee that their electoral power would not be further reduced. Since § 5 of the Act is likewise free of any language qualifying or limiting the terms of abridgment which it shares with the Amendment, abridgment under § 5 presumably covers any vote dilution, not retrogression alone, and no redistricting scheme should receive preclearance without a showing that it is nondilutive. See Bossier Parish I, supra, at 493 (BREYER, J., concurring in part and concurring in judgment) (use in § 5 of Fifteenth Amendment language indicates that § 5 prohibits new plans with dilutive purposes). Such, in fact, was apparently just what Congress had in mind when it addressed § 5 to the agility of covered jurisdictions in keeping one step ahead of dilution challenges under the Constitution (and previous versions of the Voting Rights Act) by adopting successive voting schemes, each with a distinctive feature that perpetuated the abridgment of the minority vote:

"Congress had found that case-by-case litigation was inadequate to combat widespread and persistent discrimination in voting, because of the inordinate amount of time and energy required to overcome the obstructionist tactics invariably encountered in these lawsuits. After enduring nearly a century of systematic resistance to the Fifteenth Amendment, Congress might well decide to shift the advantage of time and inertia from the perpetrators of the evil to its victims." South Carolina v. Katzenbach, 383 U.S., at 328 (footnote omitted).

This evil in Congress' sights was discrimination, abridgment of the right to vote, not merely discrimination that happens to cause retrogression, and Congress' intent to frustrate the unconstitutional evil by barring a replacement scheme of discrimination from being put into effect was not confined to any one subset of discriminatory schemes. The School Board's purpose thus seems to lie at the very center of what Congress meant to counter by requiring preclearance, and the Court's holding that any nonretrogressive purpose survives § 5 is an exceedingly odd conclusion.

B

The majority purports to shoulder its burden to justify a limited reading of "abridging" by offering an argument from the "context" of § 5. Since § 5 covers only changes in voting practices, this fact is said to be a reason to think that "abridging" as used in the statute is narrower than its cognate in the Fifteenth Amendment, which covers both changes and continuing systems. Ante, at 8, 12-13. In other words, on the majority's reading, the baseline in a § 5 challenge is the status quo that is to be changed, while the baseline in a Fifteenth Amendment challenge (or one under § 2 of the Voting Rights Act) is a nondiscriminatory regime, whether extant or not. From the fact that § 5 applies only when a voting change is proposed, however, it does not follow that the baseline of abridgment is the status quo; Congress could perfectly well have decided that when a jurisdiction is forced to change its voting scheme (because of malapportionment shown by a new census, say), it ought to show that the replacement is constitutional. This, of course, is just what the unqualified language and its Fifteenth Amendment parallel would suggest.

In fact, the majority's principal reason for reading intent to abridge as covering only intent to cause retrogression is not the peculiar context of changes in the law, but Beer v. United States, 425 U.S. 130 (1976), which limited the sort of "effect" that would be an abridgment to retrogressive effect. The strength of the majority's position, then, depends on the need for parallel limitations on the purpose and effect prongs of § 5. The need, however, is very much to the contrary.

1

Insofar as Beer is authority for defining the "effect" of a redistricting plan that would bar preclearance under § 5, I will of course respect it as precedent. The policy of stare decisis is at its most powerful in statutory interpretation (which Congress is always free to supersede with new legislation), see Hilton v. South Carolina Public Railways Comm'n, 502 U.S. 197, 202 (1991), and § 5 presents no exception to the rule that when statutory language is construed it should stay construed. But it is another thing entirely to ignore error in extending discredited reasoning to previously unspoiled statutory provisions. That, however, is just what the Court does in extending Beer from § 5 effects to § 5 purpose.

Beer was wrongly decided, and its error should not be compounded in derogation of clear text and equally clear congressional purpose. The provision in § 5 barring preclearance of a districting plan portending an abridging effect is unconditional (and just as uncompromising as the bar to plans resting on a purpose to abridge). The Beer Court nonetheless sought to justify the imposition of a nontextual limitation on the forbidden abridging effect to retrogression by relying on a single fragment of legislative history, a statement from a House Report that § 5 would prevent covered jurisdictions from "'undo[ing] or defeat[ing] the rights recently won'" by blacks. Beer, supra, at 140 (quoting H.R. Rep. No. 91-397, p. 8 (1969)). Relying on this one statement, however, was an act of distorting selectivity, for the legislative history is replete with references to the need to block changes in voting practices that would perpetuate existing discrimination and stand in the way of truly nondiscriminatory alternatives. In the House of Representatives, the Judiciary Committee noted that "even after apparent defeat[s] resisters seek new ways and means of discriminating. Barring one contrivance too often has caused no change in result, only in methods," H.R. Rep. No. 439, 89th Cong., 1st Sess., 10 (1965), and the House Report described how jurisdictions had used changes in voting practices to stave off reform. By making trifling changes in registration requirements, for example, Dallas County, Alabama, was able to terminate litigation against it without registering more than a handful of minority voters, see id., at 10-11, and new practices were similarly effective devices for perpetuating discrimination in other jurisdictions as well, see S. Rep. No. 162, pt. 3, pp. 8-9 (1965) (Joint Statement of Individual Views by Sens. Dodd, Hart, Long, Kennedy, Bayh, Burdick, Tydings, Dirksen, Hruska, Fong, Scott, and Javits). After losing voting rights cases, jurisdictions would adopt new voting requirements "'as a means for continuing the rejection of qualified Negro applicants.'" Id., at 12 (quoting United States v. Parker, 236 F. Supp. 511, 517 (MD Ala. 1964)). Thanks to the discriminatory traditions of the jurisdictions covered by § 5, these new practices often avoided retrogression even as they stymied improvements. In the days before § 5, the ongoing litigation would become moot and minority litigants would be back at square one, shouldering the burden of new challenges with the prospect of further dodges to come. Beer, supra, at 152, n. 9 (Marshall, J., dissenting).

The intent of Congress to address the frustration of running to stay in place was manifest when it extended the Voting Rights Act in 1969:

"Prior to the enactment of the 1965 act, new voting rules of various kinds were resorted to in several States in order to perpetuate discrimination in the face of

adverse Federal court decrees and enactments by the Congress. . . . In order to preclude such future State or local circumvention of the remedies and policies of the 1965 act, [§ 5 was enacted]. . . .

"The record before the committee indicates that as Negro voter registration has increased under the Voting Rights Act, several jurisdictions have undertaken new, unlawful ways to diminish the Negroes' franchise and to defeat Negro and Negro-supported candidates. The U.S. Commission on Civil Rights has reported that these measures have taken the form of switching to at-large elections where Negro voting strength is concentrated in particular election districts and facilitating the consolidation of predominently [sic] Negro and predominently [sic] white counties. Other changes in rules or practices affecting voting have included increasing filing fees in elections where Negro candidates were running; abolishing or making appointive offices sought by Negro candidates; extending the term of office of incumbent white officials, and withholding information about qualifying for office from Negro candidates." H.R. Rep. No. 91-397, at 6-7.

See also 115 Cong. Rec. 38486 (1969) (remarks of Rep. McCulloch) (listing "new methods by which the South achieves an old goal" of maintaining white control of the political process).

Congress again expressed its views in 1975:

"In recent years the importance of [§ 5] has become widely recognized as a means of promoting and preserving minority political gains in covered jurisdictions. . . .

. . . . .

". . . As registration and voting of minority citizens increases, other measures may be resorted to which would dilute increasing minority voting strength. Such other measures may include switching to at-large

elections, annexations of predominantly white areas, or the adoption of discriminatory redistricting plans." S. Rep. No. 94-295, pp. 15-17 (citation omitted).

Congress thus referred to § 5 as a way to make the situation better ("promoting"), not merely as a stopgap to keep it from getting worse ("preserving").

It is all the more difficult to understand how the majority in Beer could have been so oblivious to this clear congressional objective, when a decade before Beer the Court had realized that modifying legal requirements was the way discriminatory jurisdictions stayed one jump ahead of the Constitution. In United States v. Mississippi, 380 U.S. 128 (1965), the Court described a series of ingenious devices preventing minority registration, and in South Carolina v. Katzenbach, 383 U.S. 301 (1966), the Court said that

"Congress knew that some of the States . . . had resorted to the extraordinary stratagem of contriving new rules of various kinds for the sole purpose of perpetuating voting discrimination in the face of adverse federal court decrees. Congress had reason to suppose that these States might try similar maneuvers in the future in order to evade the remedies for voting discrimination contained in the Act itself." Id., at 335 (footnote omitted); see also id., at 314-315.

Likewise, well before Beer, our nascent dilution jurisprudence addressed practices mentioned in the congressional lists of tactics targeted by § 5. See, e.g., White v. Regester, 412 U.S. 755, 765-766, 768-769 (1973).

In fine, the full legislative history shows beyond any doubt just what the unqualified text of § 5 provides. The statute contains no reservation in favor of customary abridgment grown familiar after years of relentless discrimination, and the preclearance requirement was not enacted to authorize covered jurisdictions to pour old poison into new bottles. See post, at 2-3 (BREYER, J., dissenting). Beer was wrong, and while it is entitled to stand under our traditional stare decisis in statutory interpretation, stare decisis does not excuse today's decision to compound Beer's error.

2

Giving purpose-to-abridge the broader, intended reading while preserving the erroneously truncated interpretation of effect would not even result in a facially irrational scheme. This is so because [intent to dilute is conceptually simple, whereas a dilutive abridgment-in-fact is not readily defined and identified independently of dilutive intent.] A purpose to dilute simply means to subordinate minority voting power; exact calibration is unnecessary to identify what is intended. Any purpose to give less weight to minority participation in the electoral process than to majority participation is a purpose to discriminate and thus to "abridge" the right to vote. No further baseline is needed because the enquiry goes to the direction of the majority's aim, without reference to details of the existing system.

Dilutive effect, for the reason the majority points out, is different. Dilutive effect requires a baseline against which to compare a proposed change. While the baseline is in theory the electoral effectiveness of majority voters, dilution is not merely a lack of proportional representation, see Davis v. Bandemer, 478 U.S. 109, 131 (1986) (opinion of White, J.), and we have held that the maximum number of possible majority-minority districts cannot be the standard, see, e.g., Miller v. Johnson, 515 U.S., at 925-926. Thus we have held that an enquiry into dilutive effect must rest on some idea of a reasonable allocation of power between minority and majority voters; this requires a court to compare a challenged voting practice with a reasonable alternative practice. See Holder v. Hall, 512 U.S. 874, 880 (1994) (opinion of KENNEDY, J.); id., at 887-888 (O'CONNOR, J., concurring in part and concurring in judgment); see also Johnson v. De Grandy, 512 U.S. 997, 1018 (1994). Looking only to retrogression in effect, while looking to any dilutive or other abridgment in purpose, avoids the difficulty of baseline derivation. The distinction was not intended by Congress, but such a distinction is not irrational.

Indeed, the Justice Department has always taken the position that Beer is limited to the effect prong and puts no limitation on discriminatory purpose in § 5. See Brief for Federal Appellant 32-33. [The Justice Department's longstanding practice of refusing to preclear changes that it determined to have an unconstitutionally discriminatory purpose, both before and after Beer,] is entitled to "particular deference" in light of the Department's "central role" in administering § 5. Dougherty County Bd. of Ed. v. White, 439 U.S. 32, 39 (1978); see also United States v. Sheffield Bd. of Comm'rs, 435 U.S. 110, 131-132 (1978); Perkins v. Matthews, 400 U.S. 379, 390-391 (1971). Most significant here, the fact that the Justice Department has for decades understood Beer to be limited to effect demonstrates that such a position is entirely consistent and coherent with the law as declared in Beer, even though it may not have been what Congress intended.

3

Giving wider scope to purpose than to effect under § 5 would not only preserve the capacity of § 5 to bar preclearance to all intended violations of the Fifteenth Amendment, it would also enjoy the virtue of consistency with prior decisions apart from Beer. In Richmond v. United States, 422 U.S. 358 (1975), the Court held that a city's territorial annexation reducing the percentage of black voters could not be recognized as a legal wrong under the effect prong of § 5, but remanded for further consideration of discriminatory purpose. The majority distinguishes Richmond as "nothing more than an ex necessitate limitation upon the effect prong in the particular context of annexation." Ante, at 9. But in fact, Richmond laid down no eccentric effect rule and is squarely at odds with the majority's position that only an act taken with intent to produce a forbidden effect is forbidden under the intent prong.

As to forbidden effect, the Richmond Court said this:

"As long as the ward system fairly reflects the strength of the Negro community as it exists after the annexation, we cannot hold, without more specific legislative direction, that such an annexation is nevertheless barred by § 5. It is true that the black community, if there is racial bloc voting, will command fewer seats on the city council; and the annexation will have effected a decline

in the Negroes' relative influence in the city. But a different city council and an enlarged city are involved after the annexation. Furthermore, Negro power in the new city is not undervalued, and Negroes will not be underrepresented on the council.

"As long as this is true, we cannot hold that the effect of the annexation is to deny or abridge the right to vote." 422 U.S., at 371.

As Richmond's references to "undervaluation" and "underrepresentation" make clear, the case involves application of standard Fifteenth Amendment principles to the annexation context, not an annexation exception. As long as the postannexation city allowed black voters to participate on equal terms with white voters, the annexation did not "abridge" their voting rights even if they thereafter made up a smaller proportion of the voting population. The Court also held, however, that in adopting the very plan whose effect had been held to be outside the scope of legal wrong, the city could have acted with an unlawful, discriminatory intent that would have rendered the annexation unlawful and barred approval under § 5:

"[I]t may be asked how it could be forbidden by § 5 to have the purpose and intent of achieving only what is a perfectly legal result under that section and why we need remand for further proceedings with respect to purpose alone. The answer is plain, and we need not labor it. An official action, whether an annexation or otherwise, taken for the purpose of discriminating against Negroes on account of their race has no legitimacy at all under our Constitution or under the statute. Section 5 forbids voting changes taken with the purpose of denying the vote on the grounds of race or color." Id., at 378.

It follows from Richmond that a plan lacking any underlying purpose to cause disqualifying retrogression may be barred by a discriminatory intent.

The majority's attempt to distinguish Pleasant Grove v. United States, 479 U.S. 462 (1987), is equally vain. Whereas Richmond dealt with the argument that law and logic barred finding a disqualifying intent when effect was lawful, Pleasant Grove dealt with the argument that finding a disqualifying intent was impossible in fact. The Court in Pleasant Grove denied preclearance to an annexation that added white voters to the city's electorate, despite the fact that at the time of the annexation minority voting strength was nonexistent and officials of the city seeking the annexation were unaware of any black voters whose votes could be diluted. One thing is clear beyond peradventure: the annexation in that case could not have been intended to cause retrogression. No one could have intended to cause retrogression because no one knew of any minority voting strength from which retrogression was possible. 479 U.S., at 465, n. 2. The fact that the annexation was nonetheless barred under the purpose prong of § 5, 11 years after Beer, means that today's majority cannot hold as they do without overruling Pleasant Grove.

The majority seeks to avoid Pleasant Grove by describing it as barring "future retrogression" by nipping any such future contingency even before the bud had formed. This gymnastic, however, not only overlooks the contradiction between Pleasant Grove's holding that a voting change without possible retrogressive intent could fail under the purpose prong and the majority's reasoning today that the baseline for the purpose prong is the status quo; it even ignores what the Court actually said. While the Pleasant Grove Court said that impermissible purpose could relate to anticipated circumstances, 479 U.S., at 471-472, it said nothing about anticipated retrogression (a concept familiar to the Court since the time of Beer). The Court found it "plausible" that the city had simply acted with "the impermissible purpose of minimizing future black voting strength." 479 U.S., at 471-472 (footnote omitted). The Court spoke of "minimizing," not "causing retrogression to." But there is more:

"One means of thwarting [integration] is to provide for the growth of a monolithic white voting block, thereby effectively diluting the black vote in advance. This is just as impermissible a purpose as the dilution of present black voting strength. Cf. City of Richmond, [422 U.S.,] at 378." Id., at 472.

That is, a nonretrogressive dilutive purpose is just as impermissible under § 5 as a retrogressive one. Today's holding contradicts that. The majority is overruling Pleasant Grove.

The majority proffers no justification for denying the precedential value of Pleasant Grove. Instead it observes that reading the purpose prong of § 5 as covering more than retrogression (as Richmond and Pleasant Grove read it) would "exacerbate the 'substantial' federalism costs that the preclearance procedure already exacts." Ante, at 15. But my reading, like the Court's own prior reading, would not raise the cost of federalism one penny above what the Congress meant it to be. The behavior of Bossier Parish is a plain effort to deny the voting equality that the Constitution just as plainly guarantees. The point of § 5 is to thwart the ingenuity of the School Board's effort to stay ahead of challenges under § 2. Its object is to bring the country closer to transcending a history of intransigence to enforcement of the Fifteenth Amendment. Now, however, the promise of § 5 is substantially diminished. Now executive and judicial officers of the United States will be forced to preclear illegal and unconstitutional voting schemes patently intended to perpetuate discrimination. The appeal to federalism is no excuse. I respectfully dissent.

JUSTICE STEVENS, with whom JUSTICE GINSBURG joins, dissenting.

In its administration of the voting rights statute for the past quarter century, the Department of Justice has consistently employed a construction of the Voting Rights Act of 1965 contrary to that imposed upon the Act by the Court today. Apart from the deference such constructions are always afforded, the Department's reading points us directly to the necessary starting point of any exercise in statutory interpretation — the plain language of the statute.

It is not impossible that language alone would lead one to think that the phrase "will not have the effect" includes some temporal measure; the noun "effect" and the verb tense "will have" could imaginably give rise to a reading that requires a comparison between what is and what will be. But there is simply nothing in the word "purpose" or the entire phrase "does not have the purpose" that would lead anyone to think that Congress had anything in mind but a present-tense, intentional effort to "den[y] or abridg[e] the right to vote on account of race." See, e.g., Webster's Third New International Dictionary 1847 (1966). Ergo, if a municipality intends to deny or abridge voting rights because of race, it may not obtain preclearance.

Like JUSTICE SOUTER, I am persuaded that the dissenting opinions of Justices White and Marshall were more faithful to the intent of the Congress that enacted the Voting Rights Act of 1965 than that of the majority in Beer v. United States, 425 U.S. 130 (1976). One need not, however, disavow that precedent in order to explain my profound disagreement with the Court's holding today. The reading above makes clear that there is no necessary tension between the Beer majority's interpretation of the word "effect" in § 5 and the Department's consistent interpretation of the word "purpose." For even if retrogression is an acceptable standard for identifying prohibited effects, that assumption does not justify an interpretation of the word "purpose" that is at war with both controlling precedent and the plain meaning of the statutory text.

Accordingly, for these reasons and for those stated at greater length by JUSTICE SOUTER, I respectfully dissent.

JUSTICE BREYER, dissenting.

I agree with JUSTICE SOUTER, with one qualification. I would not reconsider the correctness of the Court's decision in Beer v. United States, 425 U.S. 130 (1976) — an "effects" case — because, regardless, § 5 of the Voting Rights Act of 1965 prohibits preclearance of a voting change that has the purpose of unconstitutionally depriving minorities of the right to vote.

As JUSTICE SOUTER points out, ante, at 360-361 opinion concurring in part and dissenting in part, Congress enacted § 5 in 1965 in part to prevent certain jurisdictions from limiting the number of black voters through "the extraordinary stratagem of contriving new rules of various kinds for the sole purpose of perpetuating voting discrimination in the face of adverse federal court decrees." South Carolina v. Katzenbach, 383 U.S. 301, 335 (1966). This "stratagem" created a moving target with a consequent risk of judicial runaround. See, e.g., Perkins v. Matthews, 400 U.S. 379, 395-396 (1971). And this "stratagem" could prove similarly effective where the State's "new rules" were intended to retrogress and where they were not. Indeed, since at the time, in certain places, historical discrimination had left the number of black voters at close to zero, retrogression would have proved virtually impossible where § 5 was needed most.

An example drawn from history makes the point clear. In Forrest County, Mississippi, as of 1962, precisely three-tenths of 1% of the voting age black population was registered to vote. United States v. Mississippi, 229 F. Supp. 925, 994, n. 86 (SD Miss. 1964) (dissenting opinion), rev'd, 380 U.S. 128 (1965). This number was due in large part to the county registrar's discriminatory application of the State's voter registration requirements. Prior to 1961, the registrar had simply refused to accept voter registration forms from black citizens. See United States v. Lynd, 301 F.2d 818, 821 (CA5 1962). After 1961, those blacks who were allowed to apply to register had been subjected to a more difficult test than whites, while whites had been offered assistance with their less taxing applications. And the registrar, upon denying the applications of black citizens, had refused to supply them with an explanation. Id., at 822. The Government attacked these practices, and the Fifth Circuit enjoined the registrar from "[f]ailing to process applications for registrations submitted by Negro applicants on the same basis as applications submitted by white applicants." Id., at 823.

Mississippi's "immediate response" to this injunction was to impose a "good moral character requirement," Mississippi, supra, at 997, a standard this Court has characterized as "an open invitation to abuse at the hands of voting officials," Katzenbach, supra, at 313. One federal judge believed that this change was designed to avoid the Fifth Circuit's injunction by "defy[ing] a Federal Appellate Court determination that particular applicants were qualified [to vote]." Mississippi, supra, at 997. Such defiance would result in maintaining — though not, in light of the absence of blacks from the Forrest County voting rolls, in increasing — white political supremacy.

This is precisely the kind of activity for which § 5 was designed, and the purpose of § 5 would have demanded its application in such a case. See, e.g., Perkins, supra, at 395-396 (Congress knew that the "Department of Justice d[id] not have the resources to police effectively all the States . . . covered by the Act," and § 5 was intended to ensure that States not institute "new laws with respect to voting that might have a racially discriminatory purpose"); Katzenbach, supra, at 314 (Prior to 1965, "[e]ven when favorable decisions ha[d] finally been obtained, some of the States affected ha[d] merely switched to discriminatory devices not covered by the federal decrees").

And nothing in the Act's language or its history suggests the contrary. See, e.g., H.R. Rep. No. 439, 89th Cong., 1st Sess., 10 (1965) ("Barring one contrivance too often has caused no change in result, only in methods"); S. Rep. No. 162, 89th Cong., 1st Sess., pt. 3, p. 12 (1965) (joint views of 12 members of Senate Judiciary Committee, describing United States v. Parker, 236 F. Supp. 511, 517 (MD Ala. 1964), in which a jurisdiction responded to an injunction by instituting various means for "the rejection of qualified Negro applicants"); Hearings on H.R. 6400 before Subcommittee No. 5 of the House Committee on the Judiciary, 89th Cong., 1st Sess., 5 (1965) (testimony of Attorney General Katzenbach) (discussing those jurisdictions that are "able, even after apparent defeat in the courts, to devise whole new methods of discrimination"); Hearings on S. 1564 before the Senate Committee on the Judiciary, 89th Cong., 1st Sess., pt. 1, p. 11 (1965) (testimony of Attorney General Katzenbach) (similar).

[It seems obvious, then, that if Mississippi had enacted its "moral character" requirement in 1966 (after enactment of the Voting Rights Act), a court applying § 5 would have found "the purpose . . . of denying or abridging the right to vote on account of race," even if Mississippi had intended to permit, say, 0.4%, rather than 0.3%, of the black voting age population of Forrest County to register.] And if so, then irrespective of the complexity surrounding the administration of an "effects" test, the answer to today's purpose question is "yes."