MEMORANDUM OPINION

COX, Circuit Judge, with whom ALBRITTON, Chief Judge, joins.

I. Introduction

The plaintiffs, all Alabama voters, have challenged their state house-of-representative and senate districts under the equal-protection principles announced in Shaw v. Reno, 509 U.S. 630, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). This court's earlier published opinions have adequately recited this action's history. In unpublished orders, we have granted summary judgment in the defendants' favor on five of those districts on various grounds: house districts (HDs) 73, 82, 88, and 101, and senate district (SD) 28. We held a trial on the remaining claims, which challenge SDs 21, 25, 29, 30, and 34 and HDs 63, 75, 86, and 89, and now enter the following findings of fact and conclusions of law.

We conclude that race predominated in the drawing of seven of the nine districts under challenge: SDs 21, 25, 29 and 30, and HDs 63, 75, and 86; that those districts do not survive strict scrutiny; and that they are thus unconstitutional. Before we reach those conclusions, however, we address the global, procedural defenses that the defendants have interposed.

II. Procedural Defenses A. Standing

The defendants first argue, as they did in support of their summary judgment motions, that the plaintiffs lack standing. The reason, according to the defendants, is that the plaintiffs' real goal is to eliminate the majority-minority districts in Alabama, and that the "primary focus" of their case is therefore not on the districts they inhabit (all of which are majority white), but the adjacent majority-black districts. Thus, in applying the analysis of United States v. Hays, which controls on the issue of standing, the court must first determine which districts the plaintiffs are really challenging.

The Court in Hays was confronted as we are with the issue of what district the plaintiffs were challenging. The Hays plaintiffs challenged the entire legislative act that created Louisiana's congressional districts. See Hays, 515 U.S. at 746, 115 S.Ct. at 2437. Applying any district-limited concept of standing required the Court, therefore, to infer from other indicia in the case — the evidence presented at trial, for instance, and the district court's fact-findings — what district at bottom was actually under challenge. See id. at 745, 115 S.Ct. at 2436. The plaintiffs here, by contrast, had the benefit of Hays's district-limited standing rule and were able to particularize their claims by district. We can therefore thank them, and the drafters of the Federal Rules of Civil Procedure, for saving us from floundering in the vagueness of the infinitely manipulable term "primary focus." Instead, we can look to the complaint. That is where the Rules say claims reside if, as here, there is no superseding pretrial order. See Fed.R.Civ.P. 8(a); Fed.R.Civ.P. 16(e). And there, plain as day, the plaintiffs challenge the districts in which they vote. They do not challenge the majority-black districts next door, as the defendants contend. They would perhaps have stronger claims if they did challenge the next-door districts, but that does not mean they must be deemed to challenge those rather than the ones they say they challenge.

That decided, we can turn to the question of whether they have standing to attack their districts of residence under Hays's interpretation of the Case or Controversy Clause. Many trees have perished in the academic pursuit of an understanding of Hays's theories of constitutional harm. See, e.g., Judith Reed, Sense and Nonsense: Standing in the Racial Districting Cases as a Window on the Supreme Court's View of the Right to Vote, 4 Mich.J. Race L. 389 (1999); Samuel Issacharoff Pamela S. Karlan, Standing and Misunderstanding in Voting Rights Law, 111 Harv.L.Rev. 2276 (1998); Melvyn R. Durchslag, United States v. Hays: An Essay on Standing to Challenge Majority-Minority Districts, 65 U.Cin.L.Rev. 341 (1997); John Hart Ely, Standing to Challenge Pro-Minority Gerrymanders, 111 Harv.L.Rev. 576 (1997); Pamela S. Karlan, Still Hazy After All These Years: Voting Rights in the Post-Shaw Era, 26 Cumb.L.Rev. 287 (1995-96). This academic wisdom suggests that applying Hays's theory of standing in a principled way to our case, which is apparently unique among the Shaw cases in that we have white plaintiffs challenging majority-white districts that are next door to engineered minority "safe seats," may be difficult.

But we are charged with applying the law as it is. Consequently, we read Hays in its simplest terms, arbitrary as that may seem to academia: a Shaw plaintiff can satisfy the first element of Article III standing, injury-in-fact, by showing (1) that he lives in the district he challenges and (2) that the district is "racially gerrymandered." This is indeed how simply the Supreme Court has applied Hays in later cases. See Bush v. Vera, 517 U.S. 952, 957-58, 116 S.Ct. 1941, 1951, 135 L.Ed.2d 248 (1996); Shaw v. Hunt, 517 U.S. 899, 904, 116 S.Ct. 1894, 1900, 135 L.Ed.2d 207 (1996). The second prong, as we read Hays, admittedly conflates Article III jurisdiction with the merits, a situation not often encountered. But that is what Hays strongly implies when it speaks of the necessity that the plaintiff reside in a "gerrymandered district." Hays, 515 U.S. at 745, 115 S.Ct. at 2436. The upshot here is that we must dismiss for want of jurisdiction all claims in which the plaintiffs have failed to prove that the district is "racially gerrymandered." That second prong of the Hays test, moreover, will have to wait until we finish discussing the merits, below. At the outset, we note simply that all the plaintiffs undisputedly inhabit their challenged district or districts, and that they are therefore halfway there on their standing showing.

The Sinkfield defendants' counsel concurred in this statement of Hays's rule at the oral argument held at the trial's conclusion. The Sinkfield defendants and State defendants argue now, however, that the plaintiffs' subjectively felt harms are those that are relevant to a determination of injury-in-fact. They point to the two constitutional harms that Hays identified — representational (poor representation because representatives of racially gerrymandered districts will see their constituency monochromatically) and stigmatic (the emotional harm from being racially classified) — and then argue that none of these plaintiffs has testified to suffering either one of these harms. Thus, for example, they argue that the plaintiff Karen Outlaw's satisfaction with her present state senator defeats any assertion of representational harm, and that the fact that she does not live near a part of the district where borders are racial makes impossible any stigmatic harm.

The Supreme Court has never demanded that a resident of a "gerrymandered district" prove with evidence that either one of these two constitutionally cognizable harms has afflicted him specifically. See Bush, 517 U.S. at 957-58, 116 S.Ct. at 1951 ("[P]laintiffs Blum and Powers are residents of District 18, plaintiffs Thomas and Vera are residents of District 29, and plaintiff Orcutt is a resident of District 30. We stated in Hays that '[w]here a plaintiff resides in a racially gerrymandered district, . . . the plaintiff has been denied equal treatment because of the legislature's reliance on racial criteria, and therefore has standing to challenge the legislature's action.' . . . . Under this rule, these plaintiffs have standing to challenge Districts 18, 29, and 30." (citation omitted; other alterations in original)); Shaw II, 517 U.S. at 904, 116 S.Ct. at 1900 ("[A] plaintiff who resides in a district which is the subject of a racial-gerrymander claim has standing to challenge the legislation which created that district, but . . . a plaintiff from outside that district lacks standing absent specific evidence that he personally has been subjected to a racial classification."). The Supreme Court has, moreover, found standing to exist for a Hispanic-surnamed plaintiff challenging a Hispanic-majority district, in which case it would be difficult to conclude (under the defendants' version of the standing rule) that the plaintiff suffered representational harm. See Bush, 517 U.S. at 957, 116 S.Ct. at 1951 (plaintiff Vera has standing to challenge pro-Hispanic gerrymander of house district 29). The defendants have cited no case that applies Hays's standing rule to require harm to be subjectively felt by the plaintiff. We therefore conclude that in the present state of the law it appears that injury-in-fact is conclusively presumed from the mere fact of residence in a gerrymandered district, independent of the plaintiff's subjective assessment of harm.

B. Laches

The defendants next argue that this action is barred by laches. It is not. The elements of laches are (1) an unreasonable delay in asserting a right or claim; (2) lack of excuse for the delay; (3) undue prejudice to the defendant. See AmBrit, Inc. v. Kraft, Inc., 812 F.2d 1531, 1545 n. 75 (11th Cir. 1986); Citibank, N.A. v. Citibanc Group, Inc., 724 F.2d 1540, 1546 (11th Cir. 1984). Two elements are missing here.

First, the plaintiffs' filing of this action in 1997 was not unreasonable, and there was thus no cognizable delay. The consent decree that adopted the Reed-Buskey Plan was not entered until August 13, 1993, a little over a year before the 1994 election. Shaw v. Reno, 509 U.S. 630, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993), which many jurists perceived to be a major departure from older districting law under the Voting Rights Act, did not issue until June 28, 1993, and at that time Shaw had many people wondering how it affected Voting Rights Act jurisprudence and what Shaw prohibited. See Symposium, The Future of Voting Rights After Shaw v. Reno, 92 Mich.L.Rev. 483 (1993). Thus, to have challenged the Reed-Buskey Plan in time for the 1994 primaries, the plaintiffs would have had to have assembled, within a few short months, an action based on a mostly unsettled area of law. Reason did not demand this of them.

After the 1994 elections had passed, it did not matter when the plaintiffs sued, as long as it was in time for the 1998 election, which was the next one regularly scheduled and the only one left before new census data became available. See Ala. Const. art. IV, § 46 ("In the year nineteen hundred and six, and in every fourth year thereafter, all the senators and representatives shall be elected."). Had the plaintiffs prevailed, in all probability they would have achieved the same results from filing in 1995 as in 1997. To assert that the plaintiffs dallied for four years, from the birth of the Shaw claim in 1993 until 1997, overlooks this circumstance.

Second, there is insufficient prejudice here, at least insofar as we are worried about the present remedial possibilities. We have already ruled from the bench that special elections in November 2000 under a new plan are off the table. As for the possibilities that remain, which we discuss in greater detail below, the defendants have if anything benefitted from the delay: the life of an injunction against use of these districts, which could inconvenience the State if any incumbent dies or resigns, will now be shorter and thus pose a lesser risk of affecting the State.

Because the defendants have failed to show either an unreasonable delay or any prejudice from it, they are not entitled to relief under any principle of laches.

C. Issue Preclusion

The last procedural bar the defendants raise is issue preclusion. The Alabama Supreme Court dismissed as moot the appeals of plaintiffs John and Camilla Rice in the parallel state-court litigation concerning districts in Lee County, and according to the defendants that conclusion is binding in this litigation on the present plaintiffs. We disagree.

An Alabama court entered the judgment, and we therefore apply Alabama's law of judgments to determine the judgment's issue-preclusive effect. See 28 U.S.C. § 1738; Allen v. McCurry, 449 U.S. 90, 95, 101 S.Ct. 411, 415, 66 L.Ed.2d 308 (1980). According to its formal statement of the rule, Alabama forbids relitigation of an issue when "(1) . . . an issue in a prior action was identical to the issue litigated in the present action; (2) . . . the issue was actually litigated in the prior action; (3) . . . resolution of the issue was necessary to the prior judgment; and (4) . . . the same parties are involved in the two actions." Smith v. Union Bank Trust Co., 653 So.2d 933, 934 (Ala. 1995). The first of these elements is unsatisfied here.

For collateral-estoppel purposes, "an issue is a single, certain and material point arising out of the allegations and contentions of the parties." Christian Television Corp. v. RCH Broadcasting, 554 So.2d 989, 992 (Ala. 1989) (emphasis added) (internal quotations omitted). The only party to raise anything akin to a mootness argument was the State, which asserted that the Rices had slept on their rights, and that their claims were thus barred by laches. In its argument that the Rices' delay would prejudice the state, the State addressed only the Rices' demand for redistricting in time for the 1998 elections. When the Alabama Supreme Court raised the mootness issue sua sponte in its December 18, 1998 opinion, moreover, it alluded only to the possibility of redistricting in time for the 1998 election, the last general election before 2000 census data became available. See Rice v. Sinkfield, 732 So.2d 993, 993 (Ala. 1998) (addressing "[t]he plaintiffs' request for modification of the 1993 consent judgment"). All the supreme court apparently decided was that the passing of the 1998 election cycle mooted a request for redistricting before post-2000 reapportionment. Nowhere in the supreme court papers that we have did the possibility of court-ordered special elections in 2000, or of enjoining future use of these districts, either in special elections or in 2002, ever come up.

The different remedial requests we have had in this action distinguish the mootness issue here from the mootness issue before the Alabama Supreme Court. A question is moot under Alabama law if from its resolution "no practical relief can follow." Rice, 732 So.2d at 993. According to the supreme court, the question of the Reed-Buskey Plan's constitutionality is moot because even if the Rices got the relief they sought — redistricting — those new, constitutional districts could vanish before the next election, and the Rices would thus never enjoy representation in a constitutionally limned district. That analysis arguably shifts when the plaintiff wants a special election or an injunction against future use of the districting plan. A special election would of course afford relief. The plaintiffs do not necessarily get full relief from an injunction against future use of the districts, but in a state with as long a history as Alabama's of dropping the reapportionment ball (a history that we describe below), the relief could be meaningful. The defendants have thus failed to show that the supreme court had before it the same issue, as defined by the parties' contentions.

The issue is different, too, in another respect. The Alabama Supreme Court was applying Alabama law. Under Alabama law, obviously, the Rices' action was moot. But it is not under the federal law that determines our jurisdiction; the U.S. Supreme Court has recently determined that a challenge under § 5 of the Voting Rights Act to a districting plan is not moot simply because the plan will not necessarily be used again before decennial reapportionment; a precleared plan may be the basis for retrogression analysis following the next reapportionment, while a nonprecleared one may not. See Reno v. Bossier Parish Sch. Bd., ___ U.S. ___, ___, 120 S.Ct. 866, 871, 145 L.Ed.2d 845 (2000). Bossier Parish is not exactly on point, of course, because we are not dealing with § 5 or retrogression analysis. But the basic principle is the same; the Reed-Buskey Plan may well, as both the 1972 and 1983 Plans have in their turn, serve as the starting point for the next reapportionment. Alabama does not wipe the slate clean from one decade to the next, and in the same way that improper preclearance may infect retrogression analysis, a plan whose constitutional defects are left unidentified and unremedied could leave its own unconstitutional mark on the next plan.

III. The Merits A. Introduction

For starters, the Supreme Court has never held that a plaintiff must show an intent to create a district of a certain racial composition in order to trigger strict scrutiny. He must, rather, show that "race was the predominant factor motivating the legislature's decision to place a significant number of voters within or without a particular district." Miller v. Johnson, 515 U.S. 900, 916, 115 S.Ct. 2475, 2488, 132 L.Ed.2d 762 (1995). The plaintiff discharges this burden by showing that the districter "subordinated traditional race-neutral districting principles, including but not limited to compactness, contiguity, and respect for political subdivisions or communities defined by actual shared interests, to racial considerations." Id., 115 S.Ct. at 2488.

"Placing a significant number of voters in the district" refers to the drawing of lines, because it is ultimately boundaries that place voters in districts. If there were any doubt on that point, first principles would allay it: From the start, the "essence" of a Shaw claim has been that "the State has used race as a basis for separating voters into districts." Id. at 911, 115 S.Ct. at 2486 (emphasis added). The act of segregation, not racial composition alone, inflicts the constitutional harm; when an intent to racially segregate overrides a respect for political subdivisions and geographical communities, it "convey[s] the message that political identity is, or should be, predominantly racial." Bush v. Vera, 517 U.S. 952, 980-81, 116 S.Ct. 1941, 1962, 135 L.Ed.2d 248 (1996).

The drawing of lines does not happen to a single equal-population district in isolation. A line drawn to achieve a certain racial composition in target district X also outlines next-door district Y; the placement of certain population in district X rather than district Y leads Y to look elsewhere for people; and so on in a series of ripple effects that diminish as you move away from the target district. These ripple effects may in the end not be racial. But close enough to "ground zero" there may be enough race-tainted ripple effects (for instance, the division of a neighborhood along racial lines that leave whites in the next-door district and blacks in the target district) that we can fairly say that race predominated in the district's peopling.

That notion of varying degrees of effect is why Miller requires the plaintiff to show that a "significant number of voters" must have been divided by race in the formation of the district. Miller, 515 U.S. at 916, 115 S.Ct. at 2488. Insignificant race-tainted line drawing would not, presumably, send the message of race identity that entails the representational and stigmatic harms that Shaw interpreted the Equal Protection Clause to combat. We thus conclude that in this case any district where the districter's racial strategy had a significant enough effect on the configuration of the challenged district, the district triggers strict scrutiny and is "racially gerrymandered" for standing purposes. Once strict scrutiny is necessary, Miller affords the defendants the opportunity to advance a compelling interest to justify the racial classification. See id. at 920, 115 S.Ct. at 2490.

Our fact-findings compel the conclusion that SDs 21, 25, 29, and 30, and HDs 63, 75, and 86 require strict scrutiny. The defendants have advanced no sufficient compelling interest to justify the race-based districting, and accordingly we conclude that these districts all violate the Equal Protection Clause. The remaining districts, SD 34 and HD 89, however, do not satisfy this strict-scrutiny standard; they are thus not "racially gerrymandered," and for the reasons we explained in the standing discussion we must dismiss for want of jurisdiction the claims challenging them. Our detailed reasons follow.

B. General Background

1. Recent Reapportionment History

The Alabama legislature has historically had difficulty with reapportionment. A fifty-year refusal to reapportion — in flagrant disregard for Alabama's 1901 constitution — ended with the Supreme Court's decision in Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964). The same three-judge district court that had heard Reynolds in the early 1960s was required, in 1972, to impose its own reapportionment plan after the legislature again failed to discharge its constitutional duty following the 1970 census. See Sims v. Amos, 336 F. Supp. 924, 940 (M.D.Ala. 1972). The Sims plan introduced several districting features that have remained in subsequent plans, including the one whose districts are challenged here. First, the court reduced the number of representatives from 106 to 105 for the convenience of being able to "nest" three house-of-representative districts in each of the 35 senate districts. See id. at 936. Second, the court rejected, as inconsistent with the U.S. Constitution, Alabama's constitutional prohibition of splitting counties into more than one senate district. See id. at 939 n. 20. Third, the court refused to impose either multimember districts (which were previously used to keep populous counties intact) or to subordinate equal-population principles to county boundaries or precinct divisions. See id. at 934, 939. The resulting plan, which the court found to have been drawn without regard to race, had a total population deviation from the ideal equal population of 2.23% among house districts and 1.39% among senate districts, thus establishing a tolerance for some population departure. But it "substantially maintain[ed]" county integrity. Id. at 938; ( see also Plfs.' Exs. 229, 230 (maps of 1972 court-imposed plan)).

The legislature did more in the 1980s, although reapportionment still required judicial intervention. The first legislative plan, enacted in October 1981, was rejected by the Department of Justice (DOJ) following submission for preclearance under § 5 of the Voting Rights Act, and the first plan's replacement came too late for the DOJ to review it properly. (It was ultimately rejected.) See Burton v. Hobbie, 543 F. Supp. 235, 236 (M.D.Ala. 1982). A three-judge district court thus permitted the regular 1982 elections to go forward with the unprecleared plan on an understanding that special elections would be held in 1983. See id. at 238. In doing so, the court rejected the plaintiffs' proposal to hold the 1982 elections under a plaintiff-drafted "Plan B." Id. at 236. In Judge Frank Johnson's words, Plan B was not "racially neutral but instead discriminate[d] against white citizens." Id. at 241. Plan B's proposed changes, according to Judge Johnson, pitted white, but not black, incumbents against one another and "packed" blacks in "safe" black districts by joining rural black populations in the Black Belt with distant urban black populations. Id.

In early 1983, following the elections, the legislature enacted a plan drafted by Dr. Joe Reed (then a member of the Montgomery City Council) that passed DOJ scrutiny. This 1983 Plan came before the Burton v. Hobbie three-judge district court on a request not to order special elections. This time, the court had nothing but praise for the Plan and its racial balance:

[The 1983 Plan] is an exemplary reapportionment plan. In the entire state there is not a single instance of dilution of the black vote. The plan conforms closely to county lines, and population variances between districts are acceptable. The plan appears to represent a genuine legislative concern for maintaining the integrity of various economic, political and racial communities. It also represents a concerted effort to minimize the number of contests between incumbents. And though [the Plan] is similar to plaintiffs' Plan B, which had been previously submitted to this Court, we find that it has been sufficiently modified to satisfy our concerns as to its racial neutrality. . . . [The Plan] violates neither the State nor the Federal Constitutions.

Following the 1990 census, the legislature made an effort to enact a new districting plan. By early 1993, there were several proposed plans in circulation. One — an amended version of the Reed-Buskey Plan under challenge here — passed the Alabama House of Representatives, but never made it out of committee in the senate; another, the Senate Committee Plan, passed the senate but made no headway in the house. In the spring of 1993, the group that we call here the Sinkfield Defendants sued the Alabama Secretary of State in state court for violation of the Equal Protection Clause because the State had failed to redistrict. The State, represented by the secretary of state, entered into a consent decree agreeing to the use of the Reed-Buskey Plan.

The State then sought preclearance from DOJ. As might be expected from the nature of § 5 preclearance, the submission letter touted the racial features of the Plan to the exclusion of all others except the one-man, one-vote principle. In particular, the letter announced the creation of new majority-black house districts in Montgomery, Mobile, the Black Belt, and the Wiregrass, and the formation of new majority-black senate districts in the Black Belt and east-central Alabama. DOJ granted preclearance, and the state court imposed the Reed-Buskey Plan for use in subsequent elections. The legislature made no further redistricting attempts.

2. The Reed-Buskey Plan

The Reed-Buskey Plan, like the 1983 Plan, was the brainchild of Dr. Joe Reed. Dr. Reed disagreed with the three-judge court that approved the 1983 Plan. According to him, the 1983 Plan was not "equitably fair," insofar as it did not achieve proportional black representation in both chambers of Alabama's legislature. It was, rather, the best that Dr. Reed and his political allies could get. The mood of the early 1990s was more receptive to Dr. Reed's goal of "equitable fairness," however, and Dr. Reed thus set out to maximize the number of black representatives and senators elected to the legislature by maximizing the number of black-majority districts.

The Reed-Buskey Plan builds on the foundations of the 1983 Plan, and the 1972 Plan before it. In most of the state, the Plan retains the same core districts and numbers, and it continues to nest three house districts within each senate district everywhere except in Calhoun and Talladega Counties. But to achieve Dr. Reed's goals with Alabama's largely biracial demographics, Dr. Reed had to take advantage of residential segregation where it occurred to separate white voters from black voters and thus maximize black-voter influence, as measured by the likelihood of electing black candidates. Dr. Reed worked with districting software that provided him with a constant display of racial data, down to the census-block level. With these tools he succeeded in nearly eliminating what some in this litigation have referred to as "integrated" districts: those composed of a near 50-50 split of black and white population, or having a large black minority. The Reed-Buskey Plan has only one senate district with a black population between 30 and 60% of the total. By contrast, the "Pokey" Plan — the 1983 Plan with 1990 census data superimposed — had five such districts. The Pokey Plan had 15 house districts that are between 30 and 60% black; the Reed-Buskey Plan has only one.

The statistics show another aspect of Dr. Reed's strategy as well. If voters split on racial lines — and Dr. Reed had to assume this for his strategy to make any sense — "packing" blacks in majority-black districts can reduce overall political influence as much as "cracking" blacks by planting them as minorities in majority-white districts. Black votes are "wasted" in supermajority districts once a threshold of control is crossed, and those votes are better off forming a smaller majority in two districts. Attaining a happy medium between packing and cracking required racial fine-tuning around the edge of all districts with a majority-black population to assure a proper mix of black citizens and white "filler people."

The statistics show that Dr. Reed did a good job. If the ideal happy-medium district has a population between 60 and 70% black, there are seven ideal senate districts under the present Plan, as compared to two under the Pokey Plan. Twenty-three of the 105 house districts are between 60% and 70% black under the present plan; the Pokey Plan had only eight such house districts. In so delicate and successful an operation, Dr. Reed could not be indifferent to the racial composition of districts bordering his targeted black-majority districts: in biracial Alabama, if Dr. Reed left too many whites next door, he would skirt too close to packing; at the same time, though, he could not "waste" any black votes in the next-door districts because the votes were needed to maximize the number of ideal black districts. In the more racially mixed western and southern parts of the state, this meant as a practical matter that Dr. Reed had to pay special attention to the racial composition of most of the districts.

The strategy sacrificed respect for local government units. Under the 1983 Plan, 14 of Alabama's 67 counties were split between senate districts. The Reed-Buskey Plan often favors census-block lines over county lines; it thus splits 26 counties between senate districts, and there is no apparent reason why some counties — such as Dale or Lowndes — needed to be cleft to satisfy the equal-population demands of the U.S. Constitution. Besides dividing the major cities, which all recent plans have had to do to achieve equal population, the Reed-Buskey Plan splits a host of small towns, many in the challenged districts. In many of the divided towns, such as Ft. Deposit, Evergreen, Abbeville, and Grove Hill, the racial composition of populations on either side of the district line differs dramatically.

There was also a cost to incumbents, perhaps explaining the Plan's lack of legislative success. Under the Reed-Buskey Plan, incumbents were pitted against one another in seven house districts and two senate districts. All of the incumbents so paired were white.

3. The Political Context

Dr. Reed is a long-time member of the Alabama Democratic Conference (ADC), which he presently chairs, and whose interests Dr. Reed represented when he drew the Reed-Buskey Plan. The ADC, along with the New South Coalition (NSC) and local groups such as the Jefferson County Citizens Coalition (JCCC), are political groups organized to advance black political interests. The ADC has almost no white involvement. This lack of involvement is low enough that one of the county-level members in Tuscaloosa believed that the ADC did not permit whites to vote, and Dr. Jerome Gray, the ADC's state field director for the past 20 years, could believe (admittedly in error) that the ADC would permit whites to participate as supporting members only and not as active members.

The ADC is well-organized and influential. Its local, county-based chapters run get-out-the-vote campaigns to maximize black-voter participation; recruit and investigate candidates; distribute sample ballots marked with ADC-endorsed candidates; and provide voting assistance and monitoring on election days. In most races, ADC-endorsed candidates (who are always Democrats) carry a large majority of the black vote. In the 1994 Democratic primaries for the state legislature, ADC-endorsed candidates won a majority of the contested races for both the house and the senate. Thanks to the efforts of the ADC, the NSC, and other such groups, black-voter participation rates in Alabama are often comparable to, if not higher than, white rates.

The ADC, NSC, and other groups are well represented in the present Alabama legislature. Only one black senator is "independent"; the other seven are affiliates of either the ADC, the NSC, or the JCCC. In the house of representatives, the ADC counts twelve affiliates, the NSC an over-lapping five, and the JCCC five; there are only three "independents."

What the ADC does not have, however, is a large body of card-carrying members. It resembles the Democratic and Republican Parties in that its affiliates are mostly unknown in Alabama, which has no party-registration system; only a small cadre of activists are counted as members. It is thus impossible to know precisely where ADC supporters live without relying on proxies such as race or Democratic voting habits as reflected in election returns. In fact, the Reed-Buskey Plan could not have relied even upon Democratic voting habits to measure regional ADC membership in the electorate; voting returns are available only down to the precinct level, and in many of the challenged districts here, voting precincts were split (often on racial lines). Race demographics, on the other hand, were available at the census-block level, and Dr. Reed thus relied upon race as a proxy to measure local ADC strength.

C. The Challenged Districts

Dr. Reed's global strategy matters in this case only in the few areas of the state where the plaintiffs have challenged the districts in which they live. The plaintiff Dr. Ricardo Montiel lives in the Downs neighborhood on the south side of Tuscaloosa, and he thus votes in SD 21 and HD 63. Daniel Brown lives in the Carriage Hills subdivision of Montgomery, between the Eastern Bypass and Bell Road, and thus votes in SD 25. Genae Spinks lives in the Halcyon subdivision near Montgomery, south of I-85 and west of Taylor Road; she votes in SD 25 and HD 75. Peggy Kelley lives in Prattville, which is in SD 30. Karen Outlaw lives in Troy, and she votes in HD 89 and SD 30. Bibb Gunter lives just outside Ross Clark Circle on the west side of Dothan, and he votes in SD 29 and HD 86. Finally, Dr. Gonzalo Montiel lives in Mobile near the northeast side of the intersection of Old Shell Road and McGregor Avenue, and he thus votes in SD 34.

With the exception of an equal-population principle that permitted up to 5% deviation from exact equality, race predominated over all other factors in determining the layout of districts around Tuscaloosa, Montgomery, the central Black Belt, the Wiregrass, and Mobile. The effect on each district differed, however. The background on each district appears below, along with our conclusion on the predominance of race in each district.

1. West-Central Alabama — SD 21 and HD 63

Under the 1983 Plan, the region around Tuscaloosa was divided among three senate districts and five house districts. Old SD 21 consisted of the northern part of Hale County and the southern half of Tuscaloosa County, divided from the rest of the county, which was in SD 5, mostly along U.S. Highway 82 and the Black Warrior and Sipsey Rivers. SD 5 also included all of Pickens County. In 1990, SD 21 was about 30% black; SD 5 was only 13% black. Across the Tuscaloosa-Greene county line to the southwest lay SD 23, which extended from the Mississippi state line through the Black Belt to Lowndes County, just west of Montgomery. That swath of central Alabama is majority black, and its demographics were reflected in SD 23's 68% black population. In 1990, old SD 21 had 2354 inhabitants above the 5% overpopulation threshold adopted by Alabama's Permanent Legislative Committee on Reapportionment; old SD 5 was well within the permissible population range; and SD 23 was 7934 persons short of that range. Some redistricting would have been constitutionally required in any event.

Dr. Reed admitted that he intended to create a majority-black senate district in west Alabama. To do that he had to separate whites from blacks in the Tuscaloosa area. The Tuscaloosa area thus felt the repercussions of Dr. Reed's strategy to divide old SD 23's core counties into two majority-black senate districts. In the 1993 Plan, SD 23 preserves its middle-Black Belt core in Lowndes, Dallas, and Wilcox Counties, but it makes up for the population lost in the western Black Belt by extending tentacles south and southwest to pick up black populations in south Alabama towns such as Evergreen, Grove Hill, Monroeville, Thomasville, and Linden, all of which are split on racial lines. Meanwhile, the western part of old SD 23 (Perry, Hale, Greene, and Sumter Counties) forms the core of a new SD 24. (Old SD 24, which occupied all or part of Autauga, Bibb, and Chilton counties, does not overlap new SD 24 anywhere.) To compensate for the loss of the eastern Black Belt population that remained in old SD 23, new SD 24 reaches northward into Tuscaloosa, where it grabs populations that are black enough not to reduce the black majority already present in the new district's core counties. To get enough people in Tuscaloosa to satisfy equal protection in SD 24 without reducing its black majority, old SD 21 had to be purged of much of its black population. It could not, moreover, be permitted to take in large numbers of blacks in other areas, because that would have imperiled the global strategy; thus, SD 21 made up for its lost black population by moving west into majority-white Pickens County.

The senate-district strategy in central and west-central Alabama also drove the house redistricting in the Tuscaloosa area. Under the 1983 Plan, the Tuscaloosa area was divided among five house districts. SD 21 nested HDs 61, 62, and 63. HDs 61 and 63 occupied almost all of Tuscaloosa proper, divided among major thoroughfares such as University and McFarland Boulevards. HD 62 covered the southern and eastern parts of rural Tuscaloosa County, running over into the northern quarter of Hale County to the south of Tuscaloosa. The northern and western parts of Tuscaloosa County were shared between HD 14, which extended north into Walker County, and HD 15, which extended west to include all of Pickens County. HD 14 and 15 were nested (along with HD 13) in SD 5. Of these house districts, only HD 61 had a black majority in 1990, which was 67%. HDs 63 and 62 had black minorities of 15 and 18%, respectively. HD 15 had a white majority of 72%, while HD 14's white majority was 95%.

The shift to the Reed-Buskey Plan altered the area's house district lines dramatically, but it changed the overall regions they occupy in only one significant way: old HD 62 was deprived of southwest Tuscaloosa County. The blacker part of this area, which runs roughly along the Black Warrior River, has joined HD 72, which runs south through the Black Belt into southern Marengo County, eighty miles away. The whiter part of southwest Tuscaloosa County has joined new HD 61, which also occupies most of Pickens County and the part of Tuscaloosa County formerly occupied by HD 15. Old HD 61 changed shape and number — it is now HD 70 and is a lot less regular — but sits in the same place relative to the other house districts in the area. HD 63 retained its position and number, but it too underwent significant changes in shape. HDs 61, 62, and 63 are still nested in SD 21; HDs 70, 71, and 72 are nested in SD 24.

The extension of new HD 72 into Tuscaloosa County and the reconfiguration of HD 63 and new HD 70 serve two purposes in support of the senate districting scheme. First, they connect majority-black areas of the Black Belt to the black-majority areas of Tuscaloosa County, thus ensuring an ideal majority-black population in SD 24. Second, as Bryan Melton (the current representative for HD 70 and a representative since 1982) explained, the reconfiguration of old HD 61 into HD 70 served its own house-level purpose, as well. He discussed with Dr. Reed the problem of underpopulation in his majority-black district under the Pokey Plan. Rep. Melton also talked about the necessity of "helping his numbers" by ensuring a black majority of at least 65%; according to him, both the northern and southern eastbound protrusions from his district into HD 63, which removed black population from HD 63, were meant to achieve this result.

Dr. Reed's race-based strategy left its mark all over the borders of SDs 21 and 24 and their nested house districts. Starting at the Hale-Tuscaloosa county line, the SD 21-24 boundary travels roughly to the north along a railroad track and then through Tuscaloosa, and along the way there are no fewer than eight deviations of various sizes, all of which thread their way among census blocks to put the majority-black ones in SD 24 and the majority-white ones in SD 21. The line takes a west-bound turn in Northport, after the last of these eight irregularities (which excludes the black part of Northport from SD 21) and then travels back down to the Greene-Tuscaloosa county line. Along the way, the district boundary follows the Black Warrior, then swoops south to put a majority-white census block in SD 21, and then leaves the river again to separate a majority-white area from a majority-black area. The line between HD 63 and HD 70 splits nine voting precincts previously established by Tuscaloosa County; four of the split precincts have a black majority on one side and a white majority on the other.

Population statistics also betray the racial strategy. The part of Tuscaloosa county in SD 21 is 86% white; the SD 24 part of the county is 64% black. Within the City of Tuscaloosa, 84% of the population in SD 21 is white; only 35% of the population on the other side is. Within Tuscaloosa County, enough whites — but not too many — have been put in SD 21 to achieve the happy medium of black-population proportion between packing and cracking.

The defendants point to different reasons, none of them persuasive, for explaining each irregularity in the boundaries of these districts. Senator Phil Poole, a Democrat who represents SD 21, testified that each large irregularity had an independent reason to exist. We reject this testimony; other evidence in the case is to the contrary.

Sen. Poole's explanation for the Holt extension into both SD 21 and HD 63, which was corroborated by Rep. Melton and other witnesses, is that Sally Cook, an ADC member who lives in Holt, wanted to run for senate in a black-majority district. That may explain why Holt lies in a long northeastward extension of SD 24 into SD 21, but it does not explain why Dr. Reed divided the tiny community of Holt between majority-white and majority-black census blocks, as the demographic maps show; nor does it explain the race-tailored borders of the northeastward extension. Sen. Poole's explanation for the long eastward extension of SD 24/HD 70 down U.S. Highway 82 south of Tuscaloosa was that he suspected that several subdivisions with $250,000 houses would be developed in the then-sparsely peopled area, and that he wished to exclude these likely Republican voters from his district. That is inconsistent with SD 21's extensions to take in other affluent sprawl zones, such as the one described by the witness Curtis Travis on the west side of State Highway 63 south of Tuscaloosa. The only apparent difference between Curtis Travis's sprawl area and the one excluded from SD 21 is that the 1990 census data show one to be majority black and the other majority white. Sen. Poole explained the racial cutout into Northport only by pointing out that it had been there before, and he admitted that he would have preferred to keep in SD 21 the majority-black finger of land west of the Black Warrior River that Dr. Reed put in SD 24.

The evidence thus shows a regional race-based strategy in west-central Alabama that substantially and necessarily redrew SD 21; without excising black Tuscaloosa-area population from SD 21 to give it its squashed-butterfly shape, Dr. Reed could not have created the new SD 24. The borders of SD 21 and HD 63, moreover, wind through the most populated area in the region, separating a significant number of voters in a way explainable only by race. We accordingly conclude that both of these districts warrant strict scrutiny.

2. Montgomery Area — SD 25 and HD 75

Under the 1983 Plan, Montgomery was divided between two senate districts: 25, which was all but the northwest part of Montgomery County, and 26, which joined the northwest part of Montgomery County with the southern third or so of Elmore County. Old SD 25 was 78% white in 1990; old SD 26 was 65% black. SD 25 under the Reed-Buskey Plan has changed so much from the same-numbered district in the 1983 Plan that the parties dispute whether the 1983 Plan's district is relevant even as a comparator. The new SD 25, which now replaces old SD 26 as the county-line crosser, has two lobes: a southern lobe that consists of in-town majority-white parts of Montgomery joined to a majority-white sprawl zone east of Montgomery, and a northern lobe that includes sprawl zones in Elmore County. The two lobes are linked by a three-mile wide bridgeless strip of Tallapoosa River bottomland that includes the Montgomery city dump. This new SD 25 is 86% white, while the new SD 26, which consists of the portion of Montgomery County not occupied by SD 25, is 70% black. What has happened is a rearrangement of the two districts with a result that one is whiter, and the other blacker, than before, but we still have one majority-white senate district and one majority-black senate district.

This outcome may seem unexpected, even given Dr. Reed's stated goal of maximizing black membership in the legislature, because it effectively packs blacks in new SD 26. At the house-district level, however, there is an explanation. In 1990, the Montgomery area had two majority-black districts and four majority-white districts, two of which had a significant black minority. Under the Reed-Buskey Plan, the Montgomery area is divided into three supermajority-white house districts and three supermajority-black districts. The two former "integrated" districts were HD 76 and HD 75, which consisted respectively of southern Elmore County and the fringe-suburban and rural parts of Montgomery County. In the Reed-Buskey Plan, HD 75 has moved into southwest Elmore County, and its share of Montgomery County has shrunk to just the majority-white eastern suburban fringes of Montgomery, while HD 76 has left Elmore County completely to cover the southern and remaining eastern parts of Montgomery County. By removing the suburban fringe whites from eastern Montgomery County and joining them with white southwestern Elmore County, Dr. Reed created another supermajority-black district in Montgomery County.

The racial-demographic maps bear out this explanation for the do-si-do that SDs 25 and 26 and HDs 75 and 76 performed between the two plans. New SD 25's area of Montgomery County is conspicuously white by design: its eastern boundary runs neatly between the edge of majority-white suburbia and the mostly black rural areas around the east-Montgomery County communities of Waugh and Cecil. The western boundary of SD 25, moreover, runs from the south, dividing the whiter east half of Montgomery from the blacker west half, and then the western boundary retreats eastward as it approaches the Tallapoosa River, both to exclude from SD 25 the black neighborhoods on Montgomery's northeast side and, in a large detour, to put the Maxwell Air Force Base Gunter Annex in adjacent SD 26. On the other side of the line, in SD 26 and HD 76, Waugh, Cecil, and the rest of eastern and southern rural Montgomery County are about 52% black; this percentage is boosted by the marriage of the otherwise rural district with a little piece of inside-the-bypass Montgomery that is much more populous and 73% black.

Unlike in Tuscaloosa, the defendants presented no specific reason to explain the new Montgomery-area configuration other than Dr. Reed's max-black strategy. No one could assert that incumbents or other political aspirants were to blame: Larry Dixon, SD 25's incumbent, did not request any of the configuration of his district, which in fact according to him made his campaign more difficult because he had to familiarize himself with Elmore County. Nor was there any incumbent-protection explanation for HD 75 — its incumbent, a Democrat, lost in this district that has since elected Republicans. The defendants point to general communities of interest, but for the general reasons we explain in detail below, we cannot find that respect for nonracial communities of interest figured into this line-drawing.

Race was thus the predominant factor at work in Montgomery-area redistricting, and SD 25 and HD 75 are instrumental to the construction of a new majority-black house district in the Montgomery area because they remove and store whites who rendered old HD 75 an "integrated" district. The boundaries of these two Montgomery districts furthermore follow racial lines, determining racially where many Montgomerians vote. We therefore conclude that these two districts, like the Tuscaloosa-area districts, merit strict scrutiny.

3. Central Black Belt — SD 30

Old SD 30, which consisted of four whole counties (Butler, Crenshaw, Pike, and Dale), was a little underpopulated in 1990. Some change was necessary to comply with equal-protection demands. The two-lobed district that the Reed-Buskey Plan formed, however, is bizarre. Its southern lobe is still all of Butler, Crenshaw, and Pike Counties, but the old district ceded the part of Dale County that is south of a line apparently tailored to exclude the City of Ozark's black population. The new district's northern lobe is Autauga County and a small piece of southwestern Elmore County. Between these two parts is a 25-mile-long land bridge through eastern Lowndes County that hugs the Montgomery County line; at several points, the land bridge is well under a mile wide. No road runs through the land bridge, and thus to get from the southern part of the district to the northern part without leaving the district, one would have to walk.

Dr. Reed's statewide racial strategy neatly explains this configuration. Creating a compact district by extending the district a little to the north, into Lowndes County or southern Montgomery County (as the Senate Committee Plan did), would have disserved his global goal in two ways. First, by removing black voters from SD 23 and HD 76, it would have imperiled one of two additional majority-black districts: Lowndes County's black population was destined for the creation of two majority-black senate districts (23 and 24) in the Black Belt in place of the one that existed before (old 23), and rural Montgomery County's black population helped create majority-black HD 76, a new majority-black district in the Montgomery area. Second, the rural population of adjacent Lowndes and Montgomery Counties would not be enough to compensate for the loss of southern Dale County, which would have to be removed from SD 30 to satisfy the racial strategy for the Wiregrass, discussed in more detail below.

Joining 79% white Autauga County into SD 30 solved these problems. The incumbent Senator Wendell Mitchell, moreover, preferred having Autauga County in his district because under the 1972 Plan, when SD 30 included Lowndes and Autauga Counties, Mitchell had been successful in Autauga. The only problem left once it was decided to put Autauga in SD 30 was joining Autauga to a district that was 25 miles away at its closest point. Dr. Reed opted for a land bridge to solve that problem, and again the solution had a racial twist: the eastern edge of Lowndes County, through which the land bridge runs, is majority white, and the boundary between SD 30 and SD 23 hews to the contours of those majority-white census blocks. The SD 30 part of Lowndes County is 79% white; the rest of Lowndes County is 80% black.

Dr. Reed also gratuitously split the small south Lowndes town of Ft. Deposit (1990 population: 1240) along racial lines. The part of Ft. Deposit in SD 30 is 90% white; the rest of the town is 86% black. Both SD 30 and SD 23 are underpopulated, but SD 30 more so than SD 23. SD 30 thus needed either a piece of Ft. Deposit or something else to stay within 5% of the ideal population. But that does not explain why the divide of Ft. Deposit needed to be so neatly on racial lines, or why the 300 or so additional souls that needed to be transferred from SD 23 (or some other district) could not have been chosen geographically rather than racially.

Apart from protection of communities of interest, which we discuss below, the defendants have sought to justify this configuration only with Senator Mitchell's desire, expressed to Dr. Reed, to have Autauga County in his district. There is no reason to disbelieve this, but it does not explain the land bridge. Senator Mitchell candidly admitted that he did not express any opinion to Dr. Reed on the inclusion of Lowndes County in his district. Including all of Lowndes County and dropping Dale County altogether, rather than carving the land bridge out of eastern Lowndes County, would have created a senate district that consisted solely of intact counties and whose population would have been only 2.65% below the ideal; the present SD 30 is 4.84% below ideal. There is thus no reason other than race to explain how old SD 30's underpopulation problem was solved.

SD 30 therefore owes its configuration predominantly to race, and that race-driven configuration determined whether a significant number of people ended up in or out of the district. The district warrants strict scrutiny.

4. Wiregrass — SD 29 and HDs 86 and 89

The districting of the Wiregrass and adjacent areas changed a lot between the 1983 and 1993 Plans. The 1983 Plan's senate districts respected county lines throughout the area. Dale County was entirely in SD 30; Houston, Henry, and Geneva Counties were in SD 29; Covington, Coffee, and Escambia Counties composed SD 31; and Macon, Bullock, Russell, and Barbour Counties made up SD 28. Dale, Henry, and Houston Counties are now divided; Houston and Dale Counties are each split among three senate districts.

Because of equal-population constraints, house districts in the area were less respectful of county lines, but they were still more regular. SD 29 nested HDs 85, 86, and 87, then as now. HD 85 was all of Henry County and parts of northern and eastern Houston County. HD 86 was three-fourths of Dothan and a swath of central Houston County. HD 87 was Geneva County and a small part of western Houston County, all outside the City of Dothan. Dale County contained both HD 88, which occupied its southern two thirds, including all of Ozark, and HD 89, which comprised the northwestern third of the county and all of adjacent Pike County.

The present SD 29 comprises all of Geneva County, most of Houston County, and jagged fragments of Henry and Dale Counties. The fragments are all within HD 85, which snakes from the majority-black northeast side of Dothan up through the west side of Henry County, along the way splitting the towns of Headland and Abbeville, and thence meanders through the middle of Dale County, cleaving Ozark and Daleville. HD 86 is most of Houston County, except for the part of Dothan put in HDs 85 and 93 and the piece of western Houston County that is in HD 87. HD 89 consists of Pike County and the northern part of Dale County, including a snippet of the City of Ozark.

The only way to explain these changes is that Dr. Reed wanted to squeeze a single majority-black house district out of the white-majority Wiregrass. To do so, the demographic maps show, he strung together majority-black census blocks with narrow connecting strips. These blocks included 965 of Abbeville's 1115 blacks; 845 of Headland's 1088 blacks; and 2689 of Ozark's 3165 blacks. The portions of Houston County the district travels through are 82.6% black; the portion of Henry County is 65.5% black; and while the Dale County joints of the snake are only 28.7% black, that percentage compares to 7.7% and 11.7%, respectively, of the areas of Dale County in HD 89 and HD 93.

Even with this careful attention, because of the region's racial composition Dr. Reed arrived at a house district, HD 85, whose voting-age population is majority white, even though its total population is majority black. This difficulty of developing a black-majority district in the Wiregrass may explain the extension into Dale County, which ends up at Fort Rucker, a major military base. There was evidence to suggest that Dr. Reed believed that military bases provided ideal populations of filler people because many of the citizens living there would either vote elsewhere absentee or take little interest in local politics. (HD 85 has in fact elected a black representative in 1994 and 1998; the evidence does not tell us whether this was because whites participated at lower rates than blacks, or because the black candidate, Locy Baker, enjoyed white support.)

HD 85's nesting inside SD 29 taints SD 29 with the same racial raison d'être as HD 85 itself. Being on HD 85's periphery puts HDs 89 and 86 within the zone of effects of HD 85. HD 86 occupies more or less the same area it did under the 1983 Plan, but its border with HD 85 through Dothan — whose citizens supply the lion's share of the district's population — is racially determined; the border travels between majority-white and majority-black census blocks to divide Dothan on racial lines. The part of Dothan (whose total HD 85 and HD 86 population is 30,837) in HD 85 is 90% black; the part in HD 86 is 85% white. HD 89's border with HD 85 has a similar racially motivated look to it, especially around the City of Ozark, where the line takes a detour to include the majority-black part of Ozark in HD 85.

Dr. Reed admitted to his motivation for drawing HD 85, and the defendants do not point to any specific evidence to explain the race-based boundaries of HD 86 or HD 89. They certainly could not rely on incumbent protection or requests, as they did in other areas. The incumbent in old HD 85 was John Beasley, who lived just across the Houston-Henry county line near Columbia, a small town on the Chattahoochee River in northeast Houston County. Rep. Beasley's house was left out of the new HD 85. The Reed-Buskey Plan as originally unveiled (but not the one in the consent decree), moreover, deliberately put Rep. Beasley's house in HD 86, where his friend Joe Carothers was the incumbent: the boundary of HD 86 took one small detour to pick up three houses in southern Henry County, one of which was Rep. Beasley's.

SD 29 thus owes its bizarre shape to the nesting of HD 85, which exists for no other reason than to create a majority-black house district in the Wiregrass. HD 86, while only on the periphery of the engineering of HD 85, was significantly affected because its most populous portion — that within the City of Dothan — is divided from HD 85 on exclusively racial lines. Both of these districts thus merit strict scrutiny under the Equal Protection Clause.

HD 89, on the other hand, was affected by Dr. Reed's Wiregrass strategy only along its southern border, where HD 85 stretches to swallow some majority-black census blocks. This part of the district is sparsely populated, and there were enough other small black populations around in the region that it is hard to say that HD 85 could not have been created without these particular racial detours along HD 89's borders. We cannot conclude, therefore, that the effect of race on HD 89 was significant enough under Miller to trigger strict scrutiny.

5. Mobile — SD 34

Under the 1983 Plan, SD 34 occupied the northern and western parts of rural Mobile County. On the southern part of its eastern boundary, it had a short, wide extension into the City of Mobile. This extension stopped at I-65 (also called the "beltline"), thus including only the suburban neighborhoods outside and not the integrated in-town neighborhoods east of the beltline in Midtown Mobile. SD 34 changed little in configuration and position under the Reed-Buskey Plan. New SD 34 still includes most of rural west and northwest Mobile County, and it still has a protrusion on its southeastern side that takes in Mobile's western sprawl zones. Now, however, the protrusion has grown a jagged protuberance inside the beltline to include some majority-white census blocks in Midtown while leaving out majority-black blocks.

The racial composition of Mobile-area senate districts did not change dramatically between the 1983 and 1993 Plans. In 1983, there were two districts (32 and 34) that had very small black minorities, another (35) that was about one-quarter black, and another (33) with a black supermajority. The racial compositions of the current districts bearing those numbers are in the same ballpark under the Reed-Buskey Plan.

The change has been in the nested house districts, where Dr. Reed created a new majority-black district. In 1983, Mobile County had ten house districts: six had a large white majority; three had a large black majority; and one had a black minority of 44% of the total population. That last district, numbered 97, is gone under the present plan, which has six white-majority districts and four black districts with a black population of around 65% of the total. The old "integrated" district, HD 97, lent its number to a majority-black district whose lines generally follow those of old HD 103. Old HD 97 was bounded on the east by Ann Street, on the north by Threemile Creek (and maybe Dauphin Street for the western part of the northern boundary; we cannot tell perfectly from the small-scale maps), on the west by the beltline, and on the south by I-10 (possibly again with part of the boundary marked by other streets). That area is now divided mostly between a new HD 103 and HD 101; small pieces may be in HD 99 and the new HD 97. Most of the white part of old HD 97 has joined white population outside the beltline to form majority-white HD 101. The black part of old HD 97, most of which was in the southeast part of the district, has been matched with other majority-black neighborhoods by a narrow bridge, at one point only a narrow block wide, that snakes its way through Midtown Mobile to the north side of the in-town area. A small, majority-white neighborhood closer to downtown has ended up in the new HD 97.

All this racially motivated change affected SD 34 in two small ways: First, when the new HD 101 was nested in SD 34, it brought SD 34 a mostly white neighborhood in in-town Mobile with a racially crenelated boundary. The second racially drawn part of SD 34's perimeter is on the north side of the district's protrusion into Mobile. There, adjacent majority-black SD 33 takes a southward jag to rake in Mobile Terrace and Hillsdale Heights, two small majority-black pockets of Mobile's mostly white west-side suburbia.

The defendants offered no explanation, other than the community-of-interest evidence that we address and reject below, of the apparently race-based boundaries around the southeast edges of SD 34. As was the case in the Wiregrass and Montgomery, there is no serious incumbent-protection argument here. Mary Zoghby, the 1993 incumbent in the old "integrated" HD 97, testified credibly that I-65 was a natural dividing line between Midtown and suburban Mobile, and that the race-based division of Midtown was against her wishes.

While race thus determined the configuration of many of the Mobile County districts, SD 34 was on the edge of this racial engineering. It had to include some Midtown Mobile whites to further the creation of new HD 103, but those changes were modest in scope. The inclusion of Mobile Terrace and Hillsdale Heights in adjacent SD 33, moreover, was almost gratuitous under the global strategy because of their small population. Thus, while both these districts were affected by Dr. Reed's race-motivated districting, the effects were not significant enough to warrant strict scrutiny.

D. Nonracial Districting Criteria

1. Preserving Communities of Interest

The defendants sought to deflect the inference that race predominated in drawing many of the challenged districts by presenting testimony that all of the challenged districts were drawn to preserve the "communities defined by actual shared interests" that Miller identifies as a legitimate districting factor. Miller v. Johnson, 515 U.S. 900, 916, 115 S.Ct. 2475, 2488, 132 L.Ed.2d 762 (1995). The communities of interest varied from district to district and witness to witness. Senator Dixon, for example, considered his district, SD 25, to represent a community of interest because it was middle-class and probusiness. Senator Mitchell thought his district, SD 30, constituted a community of interest because much of it was rural, and all of the constituents were happy to see state money spent on building projects in their areas. The Reverend James Smith of Dothan sees communities of interest in SD 29's northern branches because he and other pastors may live in one county and preach in another, there are fraternal organizations that transcend county lines, and many people throughout the Wiregrass shop and seek health care in Dothan.

There are no doubt religious, class, and social communities of interest that cross county lines and whose protection might be a legitimate consideration in districting decisions. But the evidence in this case does not show that protecting communities of interest prevailed over race in the decision to put some people in one district and others in another. First, no evidence shows that Dr. Reed knew or cared about these communities of interest when he drew the district lines. Second, these communities of interest are not ones that could guide a districter's hand, even if the districter were setting out to preserve them. Religious communities permeate the state; middle-class people live everywhere; there are social organizations and outlooks that are shared by citizens scattered from the Tennessee Valley to the Wiregrass. On the other hand, seeking to preserve geographical communities of interest could determine a proper line: such communities, for example, might include voters who enroll their students in the same school system that is seeking more state funding; commuters who use the same highway that is pitted with potholes or congested; or citizens whose air is polluted by a local plant, or who want the state to subsidize a new factory in their town through tax breaks. There is nothing here to suggest that this plan sought to preserve those kinds of communities of interest, and the widespread division of towns and neighborhoods on racial lines implies that Dr. Reed in fact ignored commonalities of schools, roads, and other civic services. We therefore find that protecting nonracial communities of interest was not an important objective in crafting the Plan's lines for any of the challenged districts.

2. Compactness

The parties offered objective measures and expert testimony about the absolute compactness of the challenged districts, and how they measure up against districts elsewhere. We reject this testimony. First, the reliability of the defendants' expert's testimony is undermined by its disagreement with common sense. While it may be true that some scholars have adopted objective standards that deem meandering and contorted districts like SD 30 and HD 85 to be "compact," a glance and a grasp of the English language suffice to show that they are anything but. Second, even the plaintiffs' expert testimony was of limited usefulness. As the testimony implied, different regions will have different parameters of reasonable compactness: where crooked county lines, winding highways, and meandering rivers are the dominant "natural" boundaries in an area, a district whose compactness score is objectively reprehensible may in fact reflect an honest effort at compact districting. What is more relevant than comparison to districts in other regions, therefore, would be comparison to other districts that would be drawable in the relevant region to include similar geographical communities.

We have no such objective calculations in the record before us. The best we can do is apply an eyeball test to compare the Reed-Buskey Plan districts to other districting plans based on the same data. The only one we have in the record for senate districts is the Senate Committee Plan. Its senate districts are something of a hybrid between the 1983 Plan and the Reed-Buskey Plan: SD 21 follows pretty much the shape of the 1983 Plan; SDs 25 and 26 are rearranged as in the Reed-Buskey Plan, albeit with slightly different boundaries; SD 30 has its old configuration, except that it includes a small block of southwest Montgomery County; SD 29 is still just Henry, Houston, and Geneva Counties, but minus a piece of Henry County, which ended up in SD 28. SD 34 appears to have about the same configuration as it had under the 1983 Plan, although it is impossible to tell for sure on the small Senate Committee Plan map. These districts, apart from SD 25, have fewer irregularities in their shape, and this suggests that a districter more concerned with compactness than race would have arrived at a different district configuration from what we see in the Reed-Buskey Plan. Certainly there is no superiority in compactness in the Reed-Buskey Plan to suggest that compactness was more important than race, whose fingerprints are everywhere.

We can reach a similar conclusion based on the only alternative house districting plan we have: the House Committee Plan, on which the house never voted. Our map is small-scale, and thus hard to read, but in the areas where the presently challenged districts are located, there was not a lot of change between the 1983 Plan and this Plan. Southern Tuscaloosa County was still entirely within HD 62; HD 75 still occupied rural southern Montgomery County; and HD 86 continued to occupy central Houston County.

E. Compelling Interests

The defendants do not admit that the districts challenged here require strict scrutiny, and they have thus not explicitly advanced any countervailing compelling interest. Cf. Miller, 515 U.S. at 920, 115 S.Ct. at 2490. The Sinkfield defendants do, however, seek to justify the right of members of the ADC and similar organizations to achieve political power through racial gerrymandering. The only apparent relevance of this argument is as a possible compelling interest, and we treat it as such.

As it turns out, whether this associational interest is a compelling state interest is a question for another day, because for two reasons there is no record support for the argument. First, no evidence supports a conclusion that any other districting plan under consideration in 1993 would have violated the associational rights of ADC members, and thus that the State would have had a compelling interest to consent to the Reed-Buskey Plan to avoid violating those rights. Second, there is no way that the Plan could have been narrowly tailored to protect the associational rights of the ADC and similar organizations because Dr. Reed simply did not have the data he would have needed to protect those organizations; he was forced instead to use the overbroad proxy of race. We thus reject the defendants' apparent compelling-interest defense.

F. Remedies

As we announced from the bench, the court considers it too late, and the burden on the State too great, to direct that the State redistrict and conduct a special November 2000 legislative election in tandem with the presidential and congressional elections. The plaintiffs have asked in the alternative for a host of other remedies. Equitable remedies in this area must, as always, be tempered with a weighing of the relative burdens imposed, and they must as well focus on the constitutional violation that has been proven. See Reynolds v. Sims, 377 U.S. 533, 585, 84 S.Ct. 1362, 1393-94, 12 L.Ed.2d 506 (1964). Weighing the equities, and keeping in mind the narrow constitutional violation proven here, we grant some of the plaintiffs' requested remedies and deny others.

First, we agree with the plaintiffs that they are due an injunction against future use of the unconstitutional districts, even for special elections. No less a remedy will be effective, especially given Alabama's long-proven inability to redistrict itself in a timely fashion, and the risk that it will perpetuate the use of these unconstitutional districts. That said, the State has a legitimate interest in avoiding the expense of redistricting for a single special election in these districts, should one become necessary because of an unexpected vacancy. Therefore, in the event a special election should become necessary in any of the unconstitutional districts, the court will at that time hear from the parties, consider all the circumstances then existing, and determine the appropriate relief.

Second, we reject the plaintiffs' request that we enjoin the State from relying on the Reed-Buskey Plan as a baseline to measure retrogression under § 5 of the Voting Rights Act. While we share the plaintiffs' implicit doubt that any plan that is constitutional under Shaw could avoid retrogressing from the unconstitutional Reed-Buskey Plan, we reject this request for two reasons. The first reason is that the plaintiffs do not explain how so enjoining the State will prevent further harm from use of the unconstitutional districts. If the Southern redistricting battles of the 1990s have taught us anything, it is that the states covered by § 5 have little control over their districting decisions, even to the extent of being permitted to observe constitutional restrictions on districting. Even if Alabama wished to (or was ordered to) avoid Shaw problems by using the 1983 Plan as a baseline, DOJ could disagree, deny preclearance, and pressure Alabama into adopting a plan that (relying on the Reed-Buskey Plan as a baseline) violates the Equal Protection Clause.

The second reason is that this proposed remedy would be overbroad, because we have found only a handful of majority-white districts to be unconstitutional. Retrogression analysis will concern majority-black districts, which we could not (lacking jurisdiction) have found to be unconstitutional. Preserving these districts could have ripple effects on the districts the plaintiffs live in, again perhaps rendering them unconstitutional gerrymanders in their own right. But that is getting into the realm of speculation, and the plaintiffs may of course seek relief under Shaw following the next redistricting.

Third, the plaintiffs ask us to establish a deadline for the State to enact a precleared plan, and if the deadline is not met, to draft our own interim plan. We think that this would be unnecessarily intrusive in the State's affairs. Because we will enjoin the State from future use of the unconstitutional districts in the Reed-Buskey Plan, the State must redistrict before 2002 to avoid running afoul of that order. If it appears at any point that the State is incapable of complying with that order, the court can take the matter up then.

Finally, the plaintiffs ask for their attorney fees. Having partially prevailed, they are entitled to some fee. See Farrar v. Hobby, 506 U.S. 103, 111-12, 113 S.Ct. 566, 573, 121 L.Ed.2d 494 (1992). We encourage the parties to agree to the amount of a reasonable fee in light of our rulings, and if necessary the court will take the matter up on motion.

IV. Conclusion

For the foregoing reasons, the court concludes that SDs 21, 25, 29, and 30 and HDs 63, 75, and 86 violate the Equal Protection Clause. HD 89 and SD 34 do not. The court will enter judgment, consistent with this opinion, enjoining the State from conducting any legislative election, regular or special, in the unconstitutional districts without the prior approval of this court. The judgment will dismiss the claims challenging HD 89 and SD 34, as well as all districts for which summary judgment was granted, for want of jurisdiction.