The defendants appeal the district court's finding that the conduct of judicial elections for positions on the Circuit Court, Criminal Court, Chancery Court, and General Sessions Court in Hamilton County, Tennessee, violates Section 2 of the Voting Rights Act. See 42 U.S.C. § 1973(b) (1994) (hereinafter "Section 2"). Specifically, the district court held that the black plaintiffs made out a case of vote dilution under the three part "results" test enunciated by the Supreme Court in Thornburg v. Gingles, 478 U.S. 30, 50-51, 106 S. Ct. 2752, 92 L. Ed. 2d 25 (1986), and also under the totality of the circumstances test drawn from the nine factors identified in the Senate Report accompanying the 1982 Amendments to the Act. See S.Rep. No. 417, 97th Cong., 2nd Sess. 28-29 (1982) (hereinafter "Senate Report"). For the reasons indicated, we REVERSE.

At the time this lawsuit was filed, the Hamilton County judiciary consisted of four Circuit Court judges, three Criminal Court judges, two Chancery Court judges, and three General Sessions Court judges. After the district court handed down its opinion in this case, the Tennessee legislature passed a measure expanding the General Sessions Court in Hamilton County to five judges, legislation which was subsequently signed by Governor Don Sundquist and approved in a countywide referendum by the voters of Hamilton County. All of these judicial offices are elected at-large by the qualified voters of Hamilton County, and the elected judges serve eight-year terms. Tenn. Const. art. VI, § 4; Tenn.Code Ann. § 17-1-103 (1994). Except for the elections for the two recently-added General Sessions judges, which are to be nonpartisan, Tenn. H.B. 3273 § 1(c), the elections for these positions are partisan. Candidates for Hamilton County judicial offices run for separately designated positions, with the candidate receiving the highest number of votes declared the winner. Tenn.Code Ann. § 2-8-110 (1997 Supp.). In addition, Hamilton County judges must be members of the Tennessee Bar. Tenn.Code Ann. § 17-1-106 (1994). Under the existing system, no black lawyer has ever run for a position on the Circuit Court, Criminal Court, Chancery Court, or General Sessions Court in Hamilton County. Neither has the Governor of Tennessee ever appointed a black judge to the Hamilton County bench under his authority to designate judges to fill vacancies in the positions at issue here.

We have previously had occasion to review the controversy underlying this case, and our opinion in that matter was reported sub nom. Cousin v. McWherter at 46 F.3d 568 (6th Cir.1995). In that case, as in this one, we considered the district court's finding of a Section 2 violation. We held that the district court had failed to provide us with sufficiently detailed bases for its reasoning. Id. at 574 (noting that "we require a particularly definite record for voting rights cases") (citing Velasquez v. City of Abilene, 725 F.2d 1017, 1020 (5th Cir.1984)). Specifically, we faulted the district court for analyzing the plaintiffs' claims under "an over-arching 'totality of the circumstances' " test and not clearly addressing the application of the Gingles pre-conditions to the claims. Id. at 575. We also found that the district court failed to weigh the state's interest in "linkage"--the identity of the jurisdictional and electoral bases of its judges--as a separate and legitimate factor to be considered as part of the totality of the circumstances, id. at 576, and erred in concluding that this interest was " 'nebulous at best.' " Id. at 577. We recognized that the state's interest was but one factor in the totality of the circumstances test, but held that the linkage interest was a substantial one. Id. (citing League of United Latin Amer. Citizens v. Clements, 999 F.2d 831 (5th Cir.1993) (en banc) (hereinafter "LULAC ")). See also Houston Lawyers' Ass'n v. Attorney General of Texas, 501 U.S. 419, 426-27, 111 S. Ct. 2376, 115 L. Ed. 2d 379 (1991) (finding a state's linkage interest to be a "legitimate factor" among the totality of the circumstances). Accordingly, we directed that, on remand, "the plaintiffs must produce evidence supporting the dilution claim sufficient to carry their burden of outweighing the state's interest." 46 F.3d at 577. We expressed no opinion on the merits of the case, but vacated the district court's decision and remanded for more specific findings:

On remand, the district court is to determine the presence or absence of the three Gingles pre-conditions which plaintiffs must necessarily prove in order to establish their vote dilution claim; and if the court finds those preconditions do exist, to consider the totality of the circumstances in order to determine whether, in the context of all those circumstances, a Section 2 violation has occurred.

Id.

The district court's opinion on remand is now before us. See Cousin v. McWherter, 904 F. Supp. 686 (E.D.Tenn.1995). The district court found that the plaintiffs had met the Gingles pre-conditions, and that the totality of the circumstances weighed in favor of finding Section 2 liability. The district court ordered the State of Tennessee to submit a new plan for electing Hamilton County judges within 90 days of December 27, 1995, the date its opinion was rendered. This deadline was subsequently extended twice. Though bills attempting to designate a remedy were proposed in both the Tennessee House and Senate in January of 1996, and passed by the Judiciary Committee of the respective houses in April, 1996, neither bill gained the approval of the full body. Since the Legislature failed to propose an appropriate remedy, the district court solicited the parties' suggestions in May, 1996. In response to this invitation, the State "[took] no position as to the remedy that this Court should impose...."

The district court ultimately rejected the plaintiffs' proposed remedy--the creation of single-member districts within Hamilton County--and ordered a scheme of countywide cumulative voting for Hamilton County judgeships. Thus, under the district court's order, beginning with the August, 1998, regular election, judges for the Circuit Court, Criminal Court, Chancery Court, and General Sessions Court in Hamilton County would run in elections in which each voter is given the number of votes corresponding to the number of seats to be filled and allowed to allocate those votes among the eligible candidates as he or she sees fit. Because the district court's order encompassed any additional judgeships created in the relevant courts before the August, 1998, election, it would therefore also apply to the two newly-created General Sessions Court judgeships.

We find that the district court erred in its analysis of the Gingles pre-conditions. Indeed, our conclusion that the plaintiffs did not meet the third pre-condition would justify our reversal of this case. Even if we had found that the plaintiffs had successfully carried their Gingles burden, however, we also find that the district court erred in its application of the totality of the circumstances test. Finally, we disagree with the propriety and rationale of the plaintiffs' and the district court's proposed remedies in this case. As the district court properly recognized, single-member districting would violate Tennessee's important and substantial linkage interest. In addition, two of the three districting plans presented by the plaintiffs present additional deficiencies such that we could not approve them even if we ignored the state's linkage interest. Moreover, we consider cumulative voting a mechanism for achieving proportional representation among the judges in Hamilton County, a purpose for which Section 2 of the Voting Rights Act was not intended, and find it a particularly inappropriate remedy when applied to the election of state court judges. All three reasons underlie the result we reach in this case.

Our treatment of these issues is consistent with the latest opinions from other Courts of Appeals dealing with Voting Rights Act challenges to the conduct of judicial elections. See Milwaukee Branch of the N.A.A.C.P. v. Thompson, 116 F.3d 1194 (7th Cir.1997) (involving a challenge to the election of circuit court and state appellate court judges in Milwaukee County), cert. denied, --- U.S. ----, 118 S. Ct. 853, 139 L. Ed. 2d 753 (1998); White v. Alabama, 74 F.3d 1058 (11th Cir.1996) (involving a challenge to the method of electing judges to Alabama's civil and criminal appellate courts); Nipper v. Smith, 39 F.3d 1494 (11th Cir.1994) (en banc) (involving a challenge to Florida's method of electing judges to the state's Fourth Judicial Circuit); LULAC, 999 F.2d 831 (involving a challenge to Texas' system of electing state court judges). Each appellate court found special problems, as we do in this case, in dealing with Voting Rights Act challenges to judicial elections and with proposed remedies such as single-member districting, cumulative voting, and court expansion. See Milwaukee Branch, 116 F.3d at 1200-01 (discussing single-member districting); White, 74 F.3d at 1070-75 (reversing the imposition of a remedy increasing the size of the state appellate courts and creating a nominating commission to appoint the additional judges); Nipper, 39 F.3d at 1542-47 (addressing the possible remedies of single-member districting, cumulative voting, and the creation of a new Judicial Circuit); LULAC, 999 F.2d at 872-76 (rejecting single-member districting and limited and cumulative voting). In addition, this case involves claims substantially similar to those presented in a recent case challenging the election of Ohio state judges under the Voting Rights Act before a district court within our own Circuit. See Mallory v. State of Ohio, Case No. C-2-95-831 (S.D.Ohio) (Oct. 20, 1997).

Although this case is similar in some respects to these recent cases, it also presents a completely novel feature: the imposed remedy of cumulative voting in judicial elections. The plaintiffs conceded at oral argument that they knew of no federal case imposing this remedy in this electoral context. In fact, our research has revealed only one federal case that has ordered cumulative voting in any electoral context, and that in a case where the plaintiffs challenged the selection of a county legislature. Cane v. Worcester County, Maryland, 847 F. Supp. 369 (D.Md.1994), aff'd in part and rev'd in part, 35 F.3d 921, 928 (4th Cir.1994) (reversing the district court's imposition of cumulative voting because the district court did not adequately "give effect to the legislative policy judgments underlying the current legislative scheme"). We find it significant that the Fourth Circuit reversed the district court's imposition of cumulative voting in Cane, leaving us with no example in federal case law in which cumulative voting has been ordered and approved for elections to any office. Neither are we aware of any jurisdiction that has elected state trial judges by means of cumulative voting.

"A district court's factual findings regarding Section 2 violations and the determination of whether vote dilution has occurred are ordinarily reviewed for clear error." Cousin, 46 F.3d at 574 (citing Fed. R. Civ. Pro. 52(a) and Gingles, 478 U.S. at 79, 106 S. Ct. 2752). However, "Rule 52(a) does not inhibit an appellate court's power to correct errors of law, including those that may infect a so-called mixed finding of law and fact, or a finding of fact that is predicated on a misunderstanding of the governing rule of law." Gingles, 478 U.S. at 79, 106 S. Ct. 2752 (internal quotation marks omitted).

To meet their burden of production in an alleged Section 2 violation, the plaintiffs must satisfy three pre-conditions, enunciated by the Supreme Court in Gingles. Those conditions are:

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 .... the minority group must be able to show that it is politically cohesive .... [and 3] the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it ... usually to defeat the minority's preferred candidate.

Id. at 50-51, 106 S. Ct. 2752. Both the second and third Gingles pre-conditions thus require a court to consider the voting behavior of different races. However, the inquiry in the second pre-condition differs from that involved in the third: the former asks merely whether voters of the same race tend to vote alike, and the latter evaluates whether "a bloc-voting majority can routinely outvote" the minority, thereby "impair[ing] the ability of a protected class to elect candidates of its choice." Johnson v. DeGrandy, 512 U.S. 997, 1007, 114 S. Ct. 2647, 129 L. Ed. 2d 775 (1994) (internal citations and quotation marks omitted).

The state concedes, and uncontroverted evidence at trial showed, that blacks in Hamilton County vote sufficiently cohesively to satisfy the second Gingles pre-condition. The district court's finding to this effect is not clearly erroneous. However, we do find clear error in the district court's analysis of and conclusion concerning the third pre-condition, the extent of racial bloc voting. Since we conclude that the plaintiffs failed to meet the third pre-condition, we need not conduct a detailed inquiry into the first factor, the geographic compactness of blacks in Hamilton County, for a Section 2 claim cannot proceed unless all three Gingles pre-conditions are satisfied. See Voinovich v. Quilter, 507 U.S. 146, 158, 113 S. Ct. 1149, 122 L. Ed. 2d 500 (1993) (disposing of the plaintiffs' Section 2 claim without analyzing the first Gingles pre-condition, since the Court found that the plaintiffs had failed to meet the third pre-condition). See generally Gingles, 478 U.S. at 50, 106 S. Ct. 2752 (explaining that the circumstances contemplated by the three pre-conditions must necessarily exist "for multimember districts to operate to impair minority voters' ability to elect members of their choice"). We will review the district court's factual findings concerning racial bloc voting for clear error, and its legal conclusion that the third pre-condition has been met de novo. Cousin, 46 F.3d at 574; Gingles, 478 U.S. at 79, 106 S. Ct. 2752.

The third Gingles pre-condition requires a Section 2 plaintiff to show that the white majority votes in a manner usually to defeat the black minority's preferred candidate. 478 U.S. at 51, 106 S. Ct. 2752. Both parties in this case presented expert testimony concerning the existence and extent of racial bloc voting in Hamilton County. The district court relied heavily on the testimony of the plaintiffs' expert, Dr. Cole, see Cousin, 904 F. Supp. at 691-99 (quoting Dr. Cole at length), and on tabular data from Chattanooga City elections prepared for another case, see id. at 700-04 (reproducing Appendices A, B, and C from Brown v. Board of Comm'rs of the City of Chattanooga, 722 F. Supp. 380, 400-04 (E.D.Tenn.1989) (evaluating an unrelated Section 2 challenge to Chattanooga's practice of electing City Commissioners at-large)), in holding that the plaintiffs had met the third Gingles pre-condition. We hold that the district court erred in its reliance on both these bases and in the legal conclusions it drew from this data.

In his analysis of the issue of racial bloc voting, Dr. Cole examined elections pitting a white candidate against a black candidate. He considered black/white contests for judicial positions most relevant to this litigation, and he found five such contests: four elections for Chattanooga city judgeships between 1969 and 1991, and the 1980 Tennessee Supreme Court election. Because of the dearth of especially relevant contests, Cole also considered black/white contests for other elected positions, of which he found 29 examples between 1966 and 1993. Those additional contests included Democratic Presidential primaries; Democratic primaries for Hamilton County Council, Registrar, Public Defender, Juvenile Court Clerk, Quarterly County Court, and the Tennessee State House of Representatives; general elections for Hamilton County Trustee, Public Defender, Justice of the Peace, Constable, Quarterly County Court, and Tennessee State House of Representatives; and general elections for Chattanooga Mayor, City Council and School Board. For each election, Dr. Cole calculated the percentage of whites voting for the white candidate and the percentage of blacks voting for the black candidate, a figure said to represent racial "cohesion" in that election. Dr. Cole then averaged the cohesion figures, concluding that, for the five black/white judicial contests, the average black cohesion was 76% and the average white cohesion 80%; and for the 29 other black/white contests, the average black cohesion was 80% and the average white cohesion 88%. Dr. Cole used these averages in connection with black/white voter turnout information, determining that, in order to succeed in a Hamilton County election, a black candidate would need a number of white crossover votes exceeding the 20% average crossover suggested by his white cohesion figure. Thus, Dr. Cole concluded, a black candidate would usually lose a countywide election in Hamilton County. In addition, Dr. Cole extrapolated these figures to predict results in hypothetical elections. For example, he testified that Walter F. Williams, a black candidate who defeated a white opponent to win election to a city judgeship in 1991, would not have won the election had it been held on a countywide basis.

The methodology of the defense expert, Dr. Taebel, differed significantly from that employed by his counterpart. Dr. Taebel considered elections presenting voters with a racial choice as well as elections involving only white candidates. Dr. Taebel chose his group of elections according to: 1) their "recency," under which criterion he limited his inquiry to elections taking place in the last ten years; 2) their relevance, by which he limited himself to elections that were countywide and partisan, as are elections for the Hamilton County judgeships at issue here; and 3) their "contestedness," by which he eliminated elections in which the losing candidate received less than 15% of the vote. Using these criteria, Dr. Taebel identified 32 elections between 1982 and 1992, of which 12 involved offices requiring the winner to be a lawyer. These "lawyer-qualified" offices included the Tennessee Supreme Court; Hamilton County Circuit, Criminal, Chancery, General Sessions, and Juvenile Courts; Hamilton County District Attorney General; and Hamilton County Public Defender. The other 20 elections, termed "exogenous" elections by Dr. Taebel, involved offices such as Hamilton County Sheriff, County Executive, and Public Service Commissioner; Governor of Tennessee; and United States Senator and Representative.

Dr. Taebel tabulated the percentage of white and black votes received by each candidate in these 32 elections, a calculation which enabled him to identify the minority's preferred candidate in each. Dr. Taebel's analysis indicated that the minority's preferred candidate prevailed in seven of the 12 lawyer-qualified elections and in 16 of the 20 exogenous elections. Dr. Taebel's figures also showed that in four of the minority-preferred candidate's seven victories in the lawyer-qualified elections, and in nine of the minority-preferred candidate's 16 victories in the exogenous elections, that candidate defeated the preferred candidate of the white majority. Furthermore, those numbers also indicated that in three of the seven lawyer-qualified elections, and in seven of the 16 exogenous elections, won by the minority's preferred candidate, that candidate was also the preferred candidate of white voters. Given these conclusions, Dr. Taebel opined that the white majority did not regularly vote in such a way as to deprive black voters in Hamilton County of the opportunity to elect their preferred candidate.

We conclude that the district court's slavish adoption of Dr. Cole's analysis to support a finding that the plaintiffs met the third Gingles pre-condition was error. We reach this conclusion not merely because we disagree with Dr. Cole's methodology, but also because we find Dr. Taebel's study more relevant, and because several of Dr. Cole's data belie his conclusions. While Dr. Cole testified that, in his opinion, elections pitting two white candidates against each other offered little information for a racial bloc voting analysis, we, like Dr. Taebel, agree with the Eleventh Circuit that such elections do offer relevant information in this inquiry:

[W]e do not foreclose the consideration of electoral races involving only white candidates where the record indicates that one of the candidates was strongly preferred by black voters.... Where black voters have a genuine candidate of choice in an election involving only white candidates, then the results will be relevant to the question of whether racial bloc voting enables the white majority usually to defeat the minority's preferred candidate.

Nipper, 39 F.3d at 1540 (footnote omitted). See also DeGrandy, 512 U.S. at 1020, 114 S. Ct. 2647 (recognizing that minority voters' candidate of choice may not always share their race, but will share "common political ground" even if such a candidate does not "represent perfection to every minority voter"). Dr. Cole's limiting his analysis to black/white elections thus focuses his inquiry too narrowly. The proper inquiry is not whether white candidates do or do not usually defeat black candidates, but whether minority-preferred candidates, whatever their race, usually lose. A close look at Dr. Cole's data confirms that, even in the universe of elections he analyzed, this result does not usually occur. Black candidates won two of Dr. Cole's five black/white judicial elections, an especially impressive success rate when one considers that Timberlake, the losing black candidate in two of the three elections won by whites, was not qualified to hold the office since he was not a lawyer. In addition, blacks won 9 of Dr. Cole's 29 non-judicial black/white elections. In four of those elections--the 1974 Democratic Primary for Tennessee House District 28, the 1984 Democratic Presidential Primary, the 1990 runoff election for City Council District 7, and the 1990 general election for School Board District 7--the black candidate prevailed despite unanimous or near-unanimous white cohesion.

Moreover, the data from certain of Dr. Cole's elections reflect significant levels of white support for black candidates. For example, George Brown, the losing candidate in the 1980 election for the Tennessee Supreme Court, won 36% of the Hamilton County precincts with a voting age population that was 90% or more white. Similarly, City Judge Walter Williams won 25% of the precincts with more than 90% white voting age population in his 1991 election. Strong black candidates also occasionally attracted significant numbers of white crossover votes. As early as 1969, Bennie Harris had 31% white crossover support in his election to the position of City Judge. And Ardena Garth, the Hamilton County Public Defender, enjoyed 48% white crossover support in her 1990 election.

We reiterate that we point to these examples of black candidates' success only to show that Dr. Cole's conclusion--that black candidates in Hamilton County will usually lose for lack of sufficient white crossover support--is not borne out even by his own data. By indicating instances where black candidates have enjoyed white crossover support, we do not mean to suggest that we believe, or even that it has been shown, that black candidates are necessarily the preferred candidates of black voters. Rather, Gingles teaches that the success of minority-preferred candidates is the standard of evaluation for purposes of the third pre-condition. 478 U.S. at 51, 106 S. Ct. 2752. But cf. Clarke v. City of Cincinnati, 40 F.3d 807, 812 (6th Cir.1994) (holding that "a candidate's race can be relevant to a [Section] 2 inquiry"). Mindful of the Gingles standard, we find it significant, for example, that in three of Dr. Cole's 29 elections--the 1966 general election for Justice of the Peace, the 1968 general election for Hamilton County Trustee, and the 1972 Democratic Primary for Tennessee House District 28--the winning white candidate was also the preferred candidate of blacks. This standard also underlies our holding that the district court improperly ignored Dr. Taebel's statistics, which accurately reflected the potentiality that the minority's preferred candidate might be a white person. See Johnson, 512 U.S. at 1020, 114 S. Ct. 2647.

The district court also based its holding that the plaintiffs had met the third Gingles pre-condition on tabular data in Appendices A, B, and C of the opinion in the Brown case. See Cousin, 904 F. Supp. at 700-04 (quoting Brown, 722 F. Supp. at 400-404). In Brown, the plaintiffs successfully challenged Chattanooga's practice of electing City Commission members from the city at-large. The factual circumstances of Brown thus differ significantly from the instant case: Brown involved citywide, not countywide elections; the elected body at issue was legislative, not judicial; and the elections studied were nonpartisan, not partisan. See 722 F. Supp. at 382. These differences counsel us to review extremely carefully any conclusions reached in the Brown case. However, since the Brown Appendices formed at least a partial basis for the court's finding regarding the third pre-condition, we will review them for any light they will shed on our analysis.

These tables collected data from quadrennial Chattanooga City Commission elections from 1971 to 1987 (Appendix A), Chattanooga City judge contests from 1969 to 1987 (Appendix B), and other elections and referenda from 1970 to 1988 (Appendix C); separated the percentages of the white and black vote for each candidate; and designated each election as "Racially Polarized" or "Not Racially Polarized." See id. The Brown court found the majority of the elections in the Appendices "Racially Polarized." To the extent that the district court here relied on this designation in reaching its conclusion regarding the third Gingles pre-condition in this case, it committed clear error. The designation given the elections by the Brown court is relevant to this litigation, if at all, only in an analysis of the second Gingles pre-condition--the extent of political cohesion--a factor that is not in dispute. Viewing the Brown data through the proper perspective for an analysis of the third pre-condition, we find no basis for a finding that the white majority usually votes in a manner that denies victory to the minority's preferred candidate. In 15 of the 31 City Commission elections, and in five of the eight City Judge contests, the candidate garnering the largest percentage of the black vote also won the election. Moreover, the position supported by a majority of black voters in the 1978 referendum on the repeal of Tennessee's constitutional ban on interracial marriage, the 1978 referendum on the elected school board, and the 1984 referendum on the Metro Charter prevailed in each election. In short, the district court erred when it found the Brown data supported its finding that the plaintiffs here met the third Gingles pre-condition.

We find considerable methodological deficiency in Dr. Cole's analysis, and we also believe that neither Dr. Cole's statistics nor the election data from Brown support the district court's position. Furthermore, we believe Dr. Taebel's analysis shows that minority-preferred candidates can and do win countywide elections in Hamilton County. We therefore hold that the plaintiffs have failed to meet the third Gingles pre-condition, and accordingly we REVERSE the district court's holding to the contrary.

Because the third pre-condition is not satisfied, the plaintiffs' vote dilution claim in this case must fail. Since the claim has not passed even the initial hurdles, the plaintiffs are clearly not entitled to any remedy at all. We could, therefore, stop our inquiry, reverse the judgment, and vacate the district court's order at this point. However, mindful of the potentiality that there may be future challenges to the conduct of other lawyer-qualified elections in Hamilton County, we find it proper, and perhaps necessary, to express our views concerning the two alternative remedies proposed in this case: single-member districting, as the plaintiffs' proposed, and cumulative voting, as the district court ultimately ordered. See Milwaukee Branch, 116 F.3d at 1199 (continuing its analysis of the alleged Section 2 violation under similar circumstances).

The plaintiffs submitted three plans for dividing Hamilton County into single-member districts corresponding to the number of judgeships on a particular court: a proposed four district plan for the Circuit Court, a three district plan for the Criminal and General Sessions Courts, and a two district plan for the Chancery Court. The four district plan reflected a maximum deviation in population between the districts of 4.74%, and included one district where the black voting age population constituted a 60% majority. The three district plan had a maximum deviation in population between the districts of 10.98%, and achieved one district with a bare 50.3% majority black voting age population. Under the two district plan, blacks made up an "influence district" of 34% in one of the two districts. The districts in all three plans are contiguous and do not appear irregularly drawn; in addition, the districting plans were achieved without dividing any existing precinct. Cousin, 904 F. Supp. at 688-89.

Even if we had held plaintiffs' vote dilution claim valid, we would not have affirmed a remedy such as they proposed in this case because it is at odds with the important state interest in "linkage." Proper adherence to the principle of linkage ensures that a state court judge serves the entire jurisdiction from which he or she is elected, and that the entire electorate which will be subject to that judge's jurisdiction has the opportunity to hold him or her accountable at the polls. Single-member districts, as several courts have noted, eliminate the identity between the electoral and jurisdictional bases of its judges, thereby violating the state's significant linkage interest. Milwaukee Branch, 116 F.3d at 1201; Nipper, 39 F.3d at 1542-46; LULAC, 999 F.2d at 868-76. This linkage interest is also important because it lies at the heart of philosophical decisions about the role of judging in our system of government, a concern eloquently expressed by the Fifth Circuit in LULAC:

The decision to make jurisdiction and electoral bases coterminous is more than a decision about how to elect state judges. It is a decision of what constitutes a state court judge. Such a decision is as much a decision about the structure of the judicial office as the office's explicit qualifications such as bar membership or the age of judges. The collective voice of generations by their unswerving adherence to the principle of linkage through times of extraordinary growth and change speaks to us with power. Tradition, of course, does not make right of wrong, but we must be cautious when asked to embrace a new revelation that right has so long been wrong. There is no evidence that linkage was created and consistently maintained to stifle minority votes. Tradition speaks to us about its defining role--imparting its deep running sense that this is what judging is about.

LULAC, 999 F.2d at 872. In addition to such traditional concerns, the LULAC court correctly noted that maintaining linkage and shunning single-member districting actually favors minorities who may be concerned that their particular interests are not represented on the bench:

the subdistricting remedy sought by plaintiffs provides most judges with the same opportunity to ignore minority voters' interests without fear of political reprisal they would possess if elections were in fact dominated by racial bloc voting.

* * * * * *

After subdistricting, a handful of judges would be elected from subdistricts with a majority of minority voters. Creating safe districts would leave all but a few subdistricts stripped of nearly all minority members. The great majority of judges would be elected entirely by white voters. Minority litigants would not necessarily have their cases assigned to one of the few judges elected by minority voters. Rather the overwhelming probability would be that the minority litigant would appear before a judge who has little direct political interest in being responsive to minority concerns.

999 F.2d at 859, 873 (internal quotation marks omitted). See also Nipper, 39 F.3d at 1535 n. 79, 1543 (quoting LULAC with approval). We agree with the reasoning of the LULAC court. Similar notions motivated our holding in the previous Cousin opinion that the state's linkage interest was "legitimate" and "substantial." 46 F.3d at 577.

In addition to our philosophical aversion to the implementation of single-member districting in judicial elections, we note that the particular single-member districting plans proposed in this case present additional deficiencies. We find the plaintiffs' two district plan, which includes a district with a maximum 34% black voting age population, particularly lacking because it is based on the premise that the Section 2 violation in this case consists of an impairment of the minority's ability to influence the outcome of the election, rather than to determine it. Cousin, 904 F. Supp. at 713. As the following analysis will indicate, we would reverse any decision to allow such a claim to proceed since we do not feel that an "influence" claim is permitted under the Voting Rights Act.

The district court allowed this "influence" claim to proceed because it interpreted the legislative history of the 1982 Amendments to the Voting Rights Act to permit, without explicitly creating, such a claim. This legislative history consisted of the Senate Report's indication that the "totality of the circumstances" statutory language directs an inquiry as to whether the minority's voting strength has been " 'minimized or cancelled out,' " and the fact that the Report did not include the geographic compactness requirement, later enunciated by the Gingles Court. Cousin, 904 F. Supp. at 713 (quoting Senate Report). The district court also drew support for its position from the fact that the Gingles Court, albeit in a footnote, did not preclude an influence claim. Id. at 712-13 (citing Gingles, 478 U.S. at 46 n. 12, 106 S. Ct. 2752 n. 12).

The Supreme Court has yet to decide squarely whether Section 2 permits an influence claim. Our reading of Gingles ' footnote 12, in contrast to the district court's interpretation, reveals a Court concerned with limiting its holding, not authorizing its expansive use:

The claim we address in this opinion is one in which the plaintiffs alleged and attempted to prove that their ability to elect the representatives of their choice was impaired by the selection of a multimember electoral structure. We have no occasion to consider whether § 2 permits, and if it does, what standards should pertain to, a claim brought by a minority group that is not sufficiently large and compact to constitute a majority in a single-member district, alleging that the use of a multimember district impairs its ability to influence elections.

We also note that we have no occasion to consider whether the standards we apply to respondents' claim that multimember districts operate to dilute the vote of geographically cohesive minority groups that are large enough to constitute majorities in single-member districts and that are contained within the boundaries of the challenged multimember districts, are fully pertinent to other sorts of vote dilution claims, such as a claim alleging that the splitting of a large and geographically cohesive minority between two or more multimember or single-member districts resulted in the dilution of the minority vote.

478 U.S. at 46 n. 12, 106 S. Ct. 2752 n. 12 (emphasis in original). Later, the Court addressed a Section 2 claim alleging injury in that minority black voters lacked even a sufficient minority to influence elections, denying them the ability to elect candidates of their choice by attracting white crossover votes. Voinovich, 507 U.S. at 158, 113 S. Ct. 1149. In addressing the claim, the Voinovich Court assumed, without deciding, that an influence claim essentially similar to the one described in paragraph two of Gingles footnote 12 was actionable under Section 2. Id. at 154, 113 S. Ct. 1149. The Court recognized that if such a claim were actionable, the analysis of Gingles ' first pre-condition "would have to be modified or eliminated," but the Court did not have to take this step because it found that the plaintiffs had not satisfied the third Gingles pre-condition. Id. at 158, 113 S. Ct. 1149.

We believe the district court erred in assuming from the Gingles footnote and the Senate Report that an influence claim is actionable under Section 2. The Supreme Court's reluctance in Voinovich to state that Section 2 authorizes such a claim, when the Court was squarely presented with factual circumstances favorable to so holding, suggests that the existence of an influence cause of action should not be inferred from the Gingles footnote, which describes a case the Court has never decided, and legislative history that supports the district court's position only by suggestion. We therefore view the plaintiffs' Chancery Court-related claim as an impermissible "influence" claim, wrongly asserted under Section 2 of the Voting Rights Act. Accord McNeil v. Springfield Park Dist., 851 F.2d 937, 947-48 (7th Cir.1988) (concluding that "influence" claims undermine the purpose the Gingles pre-conditions and refusing to consider such claims). For this reason, we believe the plaintiffs proposed two district system is particularly ill-conceived.

We also find the plaintiffs' proposed three district plan deficient. Even in the one district where blacks constitute a voting age population majority, their 50.3% margin is so razor-thin that it does not meet the "safe district" standards of courts that have approved race-conscious realignments in other electoral contexts. See Latino Political Action Comm. v. City of Boston, 784 F.2d 409, 414 (1st Cir.1986) (holding that a "65 percent [majority of the general population] ... is a generally accepted threshold" where voting is racially polarized) (Breyer, J.); United Jewish Organizations of Williamsburgh, Inc. v. Carey, 430 U.S. 144, 163-64, 97 S. Ct. 996, 51 L. Ed. 2d 229 (1977) (finding it reasonable to conclude that at 65% minority population in a district is required to yield a majority of minority voting age population). Cf. McGhee v. Granville County, North Carolina, 860 F.2d 110, 113 (4th Cir.1988) (acknowledging the plaintiffs' objection to a proposed remedial plan containing a district with a 51.8% black voting age population majority since such a slim margin would give "no better than a fighting change" for a black candidate to win, but holding that the plan complied with relevant standards). In addition, the fact that the three districts exceed the permissible maximum deviation in population lends further support to the notion that this plan is particularly inappropriate as a remedy. See Brown v. Thomson, 462 U.S. 835, 842-43, 103 S. Ct. 2690, 77 L. Ed. 2d 214 (1983) (stating that a maximum deviation in population of more than 10% among legislative districts would make out a prima facie case of invidious discrimination under the Fourteenth Amendment).

We do not mean to suggest by these additional criticisms of the plaintiffs' two and three district plans that we would have approved a two or three district plan that achieved a more substantial black voting age population majority while complying with the maximum deviation figure. Nor do we wish to suggest, with respect to the three district plan, that we believe a district with a higher majority of black voting age population would necessarily elect a black candidate. We would not have approved any such a districting plan, as preceding paragraphs suggest, both because we find no vote dilution in Hamilton County judicial elections and because we disapprove of single-member districting as a remedy for judicial elections even where they violate the Voting Rights Act.

Similarly, we feel that cumulative voting, the remedy ordered by the district court in this case, is an inappropriate remedy for a Section 2 claim, and especially so when imposed on the election of state court judges.

Section 2 of the Voting Rights Act specifically precludes its use to achieve proportional representation. See 42 U.S.C. § 1973(b) ("Provided, that nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population."). See also White v. Alabama, 74 F.3d 1058, 1071-3 (11th Cir.1996) (holding that the Voting Rights Act cannot be used as a vehicle for achieving proportional representation in Alabama's appellate courts). Yet this is precisely the effect and, proponents would argue, the strength of cumulative voting as a remedy. See Lani Guinier, The Tyranny of the Majority 14-5 (1994); Pamela Karlan, Maps and Misreadings: The Role of Geographical Compactness in Racial Vote Dilution Litigation, 24 Harv. C.R.-C.L. L.Rev. 173, 231-6 (1989). The imposition of cumulative voting is thus meant to achieve an end not contemplated in the Voting Rights Act. As we indicated earlier, we have discovered only one other district court, in a case involving not judicial elections but apportionment of the county legislature, that has ordered this remedy for a Section 2 violation, a disposition subsequently reversed by the Court of Appeals. See Cane v. Worcester County, Maryland, 847 F. Supp. 369 (D.Md.1994), aff'd in part and rev'd in part, 35 F.3d 921.

The ultimate irony in this case is that even under the district court's mandated system of cumulative voting, proportional representation on the Hamilton County bench, which we believe may be the unstated goal of both the plaintiffs and the district court, is not assured. Political science teaches that the "threshold of exclusion"--the percentage of the vote necessary to guarantee the voting minority a seat--in cumulative voting equals 1 / (1 + number of seats). Karlan at 222, 232. Applying that formula to the offices at issue here, we note that the threshold of exclusion (20% for Circuit Court, 25% for Criminal Court, 33% for Chancery Court, and 16.7% for General Sessions Court) exceeds the black voting age population in Hamilton County (17%) for all the relevant offices except for the newly-created five-judge General Sessions Court. See Cousin, 904 F. Supp. at 695 (quoting Dr. Cole's testimony that the black voting age population in Hamilton County was 17% in 1990).

Cumulative voting is particularly inappropriate in judicial elections. In Hamilton County, the practical effect of this remedy would require judicial colleagues who previously ran for designated positions on the Circuit, Criminal, Chancery, and General Sessions Courts, to run against each other. This result would, predictably, undermine the treasured institution of judicial collegiality, potentially complicating the disposition of administrative matters in the Eleventh Judicial Circuit and the General Sessions Court in Hamilton County. See Nipper, 39 F.3d at 1546. In addition, forcing these judges to oppose each other would, to a substantial degree, deny them the full appreciation of the benefits of their incumbency, increasing the potential that the election would unseat some of the hardest-working and most efficient judges in the state of Tennessee. See Annual Report of the Tennessee Judiciary 1996-97 (indicating that Hamilton County Circuit and Chancery Court judges shoulder among the highest workloads in Tennessee in terms of filings and dispositions per judge). Not only would cumulative voting undermine judicial collegiality, independence, and quality, the Eleventh Circuit has explained, but

a cumulative voting system, like a subdistricting system, would encourage racial bloc voting. That, in turn, would necessarily fuel the notion that judges were influenced by race when administering justice.

The only benefit black voters could legitimately expect from a court order implementing one of the appellants' proposed remedies, which would enable them to elect black judges of their choice, is the perception that the challenged circuit and county judicial systems are colorblind....

By altering the current electoral schemes for the express purpose of electing more black judges, the federal court in fashioning the alteration, and the state courts in implementing it, would be proclaiming that race matters in the administration of justice.... Like other race-conscious remedies, this tends to entrench the very practices and stereotypes the Equal Protection Clause is set against. The case at hand, therefore, presents a remedial paradox: A remedy designed to foster a perception of fairness in the administration of justice would likely create, by the public policy statement it would make, perceptions that undermine that very ideal. In the eyes of the public and litigants, at least, justice would not remain colorblind.

Nipper, 39 F.3d at 1546 (citations, internal quotation marks, and footnote omitted). Finally, in addition to these perception concerns, "cumulative voting raises the spectre of other organized interest groups seizing control of a fraction of the state judiciary. This concern alone should caution against heralding limited and cumulative voting as panaceas for the contradictions inherent in applying section 2 to judicial elections." Mary Thrower Wickham, Note, Mapping the Morass: Application of Section 2 of the Voting Rights Act to Judicial Elections, 33 Wm. & Mary L.Rev. 1251, 1284 (1992). We share these concerns.

Even more than these concerns about perception and the prospect of a judicial branch riven by faction, we are troubled by the political theory represented in the plaintiffs' claim and the district court's opinion: that the absence of black judges on the Hamilton County bench automatically indicates a dilution of blacks' right to vote and calls for a remedy--whether single member districting, as the plaintiffs would have it, or at-large cumulative voting, as the district court ordered--that will increase the possibility of electing a black judge to a Circuit, Criminal, Chancery, or General Sessions judgeship. See Milwaukee Branch of the N.A.A.C.P. v. Thompson, 116 F.3d 1194, 1196 (7th Cir.1997) ("The possibility of increasing minority representation does not compel a jurisdiction to achieve that outcome, unless the three [Gingles ] conditions have been met and the judge is satisfied that minority voters have lacked an equal opportunity to participate in the political process."). Justice Thomas, in his eloquent concurrence in Holder v. Hall, has criticized the notion

that the purpose of the vote--or of the fully "effective" vote--is controlling seats. In other words, in an effort to develop standards for assessing claims of dilution, the Court has adopted the view that members of any numerically significant minority are denied a fully effective use of the franchise unless they are able to control seats in an elected body. Under this theory, votes that do not control a representative are essentially wasted; those who cast them go unrepresented and are just as surely disenfranchised as if they had been barred from registering. Such conclusions, of course, depend upon a certain theory of the "effective" vote, a theory that is not inherent in the concept of representative democracy itself.

512 U.S. 874, 899, 114 S. Ct. 2581, 129 L. Ed. 2d 687 (1994) (Thomas, J., concurring) (citation omitted). Justice Thomas correctly points out that "the proper apportionment of political power in a representative democracy" constitutes a "[matter] of political theory ... beyond the ordinary sphere of federal judges." Id. at 901, 114 S. Ct. 2581. Like Justice Thomas, we fear that cumulative voting is a step in the evolution of the current strategy of creating majority-minority districts to produce proportional results: "In principle, cumulative voting and other non-district-based methods of effecting proportional representation are simply more efficient and straightforward mechanisms for achieving what has already become our tacit objective: roughly proportional allocation of political power according to race." Id. at 912, 114 S. Ct. 2581.

Holder principally addressed the concept of proportional representation as it affected legislative offices. For all the reasons stated above, we find cumulative voting to be even more inappropriate when applied to judicial elections than to legislative contests. We cannot reconcile such a voting practice with the goal of the blind administration of justice, particularly since judges are not representatives who can or should solicit votes to further their political aims. Therefore, even if we found that the plaintiffs' showing met the Gingles pre-conditions or satisfied the totality of the circumstances test, we would not approve the imposition of such a remedy.

As mentioned earlier, our conclusion that the third Gingles hurdle is not met would justify our reversal of this case. However, we also find error in the legal conclusions the district court drew from its analysis of the "totality of the circumstances" test, derived from the Senate Report accompanying the Voting Rights Act and quoted in Gingles. See Senate Report at 28-29, quoted in Gingles, 478 U.S. at 36-37, 106 S. Ct. 2752. The Senate Report lists several "typical factors" that will be relevant in evaluating the totality of the circumstances in a Section 2 claim although, as the Gingles Court recognized, that list "is neither comprehensive nor exclusive," nor is it to be applied with mathematical precision. 478 U.S. at 45, 106 S. Ct. 2752 (quoting the Senate Report) (internal quotation marks and citations omitted).

We disagree with the district court's conclusion that the totality of the circumstances indicated a finding of vote dilution. Cousin, 904 F. Supp. at 712. The district court's application of this test involved a mixed question of law and fact. We therefore undertake de novo review of the district court's analysis of the totality of the circumstances, assessing each of the Senate Report's typical factors in turn, and of the district court's legal conclusion that the totality of the circumstances indicated a Section 2 violation. See Gingles, 478 U.S. at 79, 106 S. Ct. 2752 (recognizing appellate courts' power to correct errors of law infecting mixed questions of law and fact).

1. History of Discriminatory Efforts to Limit Black Voting

That Hamilton County, as part of the post-Reconstruction South, was not immune to the passage of Jim Crow laws and overt racism in the late 19th and early 20th century, as the district court found, cannot be a surprise. However, more relevant to this litigation, in our opinion, is the history behind the particular voting practice challenged here. On that score, the district court found "no proof that the establishment and maintenance of the present judicial system in Hamilton County was in any way a pretext for diluting minority voting rights." Cousin, 904 F. Supp. at 712.

Since we accept the district court's finding with respect to the origin and maintenance of at-large judicial elections in Hamilton County, we do not believe that the inglorious history of past discrimination against blacks voting rights cuts as clearly in favor of a finding of vote dilution as the district court would have it. In making this determination, we do not mean to belittle the pain caused, and the Constitutional violations represented, by the history the district court recounted. See id. at 705-06. We merely differ as to the legal significance of that history in making out a vote dilution claim in the 1990s. If anything, the record in this case suggests that the past 30 years have been marked by a complete absence of official discrimination against blacks' voting rights in Hamilton County and, as voter registration and turnout data indicate, by vigorous black participation in the democratic process.

Since the state concedes that voting is racially polarized in Hamilton County, we do not disturb the district court's finding to that effect. See id. at 706-07. As our analysis in section III indicated, however, the outcome of even a racially polarized election does not always disfavor minority voters, a result which does not support a claim of vote dilution.

3. Suspect Electoral Practices or Procedures

The district court's analysis of this factor implied that Hamilton County comprises an unusually and impermissibly large electoral district for the judicial offices at issue here, and that the size of the district enhances the opportunity for discrimination against blacks. We disagree with this premise. Hamilton County is not unusually large in comparison to other judicial districts in Tennessee. Eight other judicial districts in Tennessee are comprised of a single county, see Tenn.Code Ann. § 16-2-506 (1997 Supp.), and in every district the Circuit, Criminal, Chancery, and General Sessions Courts are elected at-large, as mandated by the Tennessee Constitution and by state statute. Tenn. Const. art. VI, § 4; Tenn.Code Ann. §§ 17-1-103 (1994). In fact, both Shelby and Davidson Counties comprise single-county judicial districts, and both counties far exceed Hamilton County in population. Apart from Hamilton County's size, there is no even allegedly discriminatory mechanism to limit or discourage black voting in Hamilton County. To the extent that the district court found that this factor indicated a Section 2 violation, therefore, we believe the court erred.

The parties agree that there is no slating process for judicial elections in Hamilton County, and there is no allegation that minorities have been denied the opportunity to run for these offices. We note only that the lack of a slating process or other obstacle to candidacy cuts against a finding of vote dilution.

5. Effects of Past Discrimination on Political Participation

We do not disturb the district court's conclusion that, as a group, blacks in Hamilton County "have been isolated from the economic and political main stream .... [and] remain a socioeconomically depressed minority with a limited ability to fund and mount political campaigns." Cousin, 904 F. Supp. at 710.

6. Overt or Subtle Racial Appeals in Political Campaigns

Although the district court made no factual finding with respect to this factor, only very isolated anecdotal evidence in record suggested that racial appeals played any part in Hamilton County's political life. Indeed, as the Brown court noted, "Chattanooga has for the most part been spared both overt and subtle racial appeals in elections." 722 F. Supp. at 396.

Although the district court concluded that very few blacks hold office in Hamilton County, we consider more relevant the extent to which blacks win when they run for office. Again, we refer to our analysis of racial bloc voting in section III, which showed that blacks in Hamilton County can and do get elected to public office.

With respect to the particular offices at issue in this litigation, the district court accurately stated that, though there are qualified blacks in Hamilton County, "no black has ever won a majority of the votes in a county-wide judicial contest." Cousin, 904 F. Supp. at 711. This assertion, however, overlooks two additional important facts. First, no black has ever run for a county judgeship, a phenomenon surely attributable at least in part to the perception that it is very difficult for a black candidate to win a countywide election. That political success is difficult, however, does not mean it is unmanageable, as the testimony of Chattanooga City Judge Walter Williams regarding the possibility of a black candidate's election to a countywide judgeship indicated. Second, evidence at trial showed that only 27 black lawyers live in Hamilton County, and three of that number already hold "lawyer-qualified" offices--City Judges Williams and Bennie Harris, and Public Defender Garth. See LULAC, 999 F.2d at 865 ("A functional analysis of the electoral system must recognize the impact of limited pools of eligible candidates on the number of minority judges that has resulted."). We do not agree that this factor clearly supports a finding of vote dilution under these circumstances.

8. Responsiveness of Elected Officials to Minority Needs

The district court's finding with respect to this factor bears quoting at length:

The Circuit Court, Chancery Court, Criminal Court and General Sessions Court Judges of Hamilton County are responsive to the particularized needs of the black community in Hamilton County inasmuch as these judges are fair and impartial in the disposition of matters before them.

Specifically, there is no proof that any of the present judges have discriminated against any African-Americans who have been involved in the judicial process as plaintiffs, defendants, attorneys, jurors, or witnesses.

Cousin, 904 F. Supp. at 711. We agree with the district court's findings regarding the responsiveness of the individual judges who currently hold Hamilton County judgeships. Indeed, we regard these findings as especially significant, for the unstated proposition underlying a claim of vote dilution is that the minority's failure to gain racial representation in the relevant office has somehow prejudiced that group. The district court's findings of fact illustrate, to the contrary, that the Hamilton County bench currently performs as the judicial branch in our system of government was meant to function.

9. Tennessee's Interest in Maintaining At-Large Elections

The district court concluded "that under the totality of the circumstances ... Tennessee's state interest in at-large elections will not suffice to overcome a violation of Section 2, because it is overcome by the substantial dilution of black voting strength that it produces in Hamilton County." Id. at 712. As the preceding subsections indicate, our analysis of the totality of the circumstances in Hamilton County does not convince us that a Section 2 violation has even occurred, much less that it is of such severity that it will overcome a state interest we still consider to be legitimate and substantial. See Cousin, 46 F.3d at 577. See also supra Section III of this opinion and cases cited therein.

We find that the district court's application of the totality of the circumstances test was clearly erroneous. We consider only the effects of past discrimination on blacks' political participation, the fifth "typical" Senate factor, to cut clearly in favor of a finding of vote dilution. However, this one factor, in combination with the others, does not outweigh the significant linkage interest Tennessee has in electing the offices at issue here on an at-large basis. Therefore, even if we had considered the plaintiffs' showing sufficient to meet the three Gingles pre-conditions, we would not have found a Section 2 violation under the totality of the circumstances.

We hold that the plaintiffs failed to show that the white majority in Hamilton County votes in such a way as usually to defeat the minority's preferred candidate, and that the plaintiffs thus failed to establish the existence of the third Gingles pre-condition. To the contrary, we believe the evidence showed that minority-preferred candidates can and do win election to office in Hamilton County, though no black candidate has ever run for one of the judgeships at issue here. This holding is sufficient to mandate our reversal of this case, even though we also find clear error in the district court's application of the totality of the circumstances test. Accordingly, we would reverse even if we had found that the plaintiffs had met the Gingles pre-conditions. Finally, we view both single-member districting and cumulative voting as inappropriate remedies in the context of judicial elections even where a Section 2 violation has been shown. Single-member districting destroys the state's substantial linkage interest in maintaining the coterminous jurisdictional and electoral boundaries of its judges. Cumulative voting, on the other hand, while it preserves the state's linkage interest, is impermissible both because it is designed to achieve purposes in conflict with the spirit of Section 2 of the Voting Rights Act and because it is a particularly inappropriate method for electing members of the judiciary. For all of these reasons, we REVERSE the district court's holdings to the contrary and VACATE its order imposing cumulative voting in the upcoming elections for Circuit Court, Criminal Court, Chancery Court, and General Sessions Court judges in Hamilton County.