This action was originally brought in October 1973 by black citizens of Gadsden County, Florida. Alleging that the at-large election system for electing school board members unconstitutionally diluted the black vote, plaintiffs sought relief under the thirteenth, fourteenth, and fifteenth amendments to the United States Constitution, as well as under 42 U.S.C. §§ 1971, 1973, 1981, and 1983. The attack on the school board election scheme was consolidated with similar challenges to the election arrangements for the Gadsden County and Quincy City Commissions. The gravamen of the complaint is that the at-large system operates to preclude the black population, which represented a substantial percentage of the registered Democratic voters in Gadsden County, from electing a member of its own race to the school board. Prior to trial on the merits, the district court dismissed the complaint as to the school board for lack of standing. On appeal, the former Fifth Circuit reversed and remanded the case for trial. McGill v. Gadsden County Commission, 535 F.2d 277, 279 (5th Cir. 1976).1

The case was tried to the district court without a jury in January 1979, and the court withheld its opinion pending the Supreme Court's decisions in Mobile v. Bolden, 446 U.S. 55, 100 S. Ct. 1490, 64 L. Ed. 2d 47 (1980), and Nevett v. Sides, 571 F.2d 209 (5th Cir. 1978), cert. denied, 446 U.S. 951, 100 S. Ct. 2916, 64 L. Ed. 2d 807 (1980). In its memorandum opinion and order, Campbell v. Gadsden County School Board, TCA No. 73-177 (N.D.Fla. Dec. 5, 1980), the district court held that the Florida primary election process for choosing school board nominees was neither conceived nor operated in a racially discriminatory fashion. Concluding that the district court's findings of fact are clearly erroneous under the standard enunciated by the Supreme Court in Rogers v. Lodge, --- U.S. ----, ----, 102 S. Ct. 3272, 3279, 73 L. Ed. 2d 1012 (1982), and Pullman-Standard v. Swint, --- U.S. ----, ----, 102 S. Ct. 1781, 1788-91, 72 L. Ed. 2d 66, 78-81 (1982), we reverse.2

Gadsden County is in the northwestern portion of Florida. In 1970, approximately 59 percent of the population and 49.36 percent of the registered voters in the county were black. As of October 1978, there were 7,662 registered white voters and 6,965 registered black voters in Gadsden County. Democrats constituted over 96 percent of all registered voters in the county.

The five members of the Gadsden County School Board are elected at-large from five residency subdistricts and serve four-year terms. Fla.Stat.Ann. §§ 230.08, .10 (West 1977).3 Both the primary and general elections are conducted at-large. In the primary, a majority of the votes cast is necessary to avoid a run-off. No such majority vote requirement exists for the general election. Fla.Stat.Ann. § 230.10 (West 1977).

The Supreme Court's decision in Rogers v. Lodge, --- U.S. ----, 102 S. Ct. 3272, 73 L. Ed. 2d 1012 (1982), provides the guidance necessary to the resolution of this case. In Lodge the Court held that "(c)ases charging that multimember districts unconstitutionally dilute the voting strength of racial minorities are ... subject to the standard of proof generally applicable to Equal Protection Clause cases." Id. at ----, 102 s.ct. At 3275 (citing Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 97 S. Ct. 555, 50 L. Ed. 2d 450 (1977); Washington v. Davis, 426 U.S. 229, 96 S. Ct. 2040, 48 L. Ed. 2d 597 (1976)). Accordingly, to prevail in a vote dilution case under the equal protection clause of the fourteenth amendment, a plaintiff must demonstrate (1) the existence of a discriminatory purpose in either the enactment or operation of the election scheme; and (2) differential impact, i.e., dilution of the minority's voting power.4

The district court ruled that the plaintiff failed to demonstrate discriminatory intent either in the enactment or in the operation of the at-large election system for school board members. At the time of the court's decision in this case, City of Mobile v. Bolden, 446 U.S. 55, 100 S. Ct. 1490, 64 L. Ed. 2d 47 (1980), was the most recent Supreme Court expression addressing the standards of proof involved in vote dilution cases. The district court found that under Bolden reliance upon the factors enumerated in Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir. 1973) (en banc), aff'd sub nom. East Carroll Parish School Board v. Marshall, 424 U.S. 636, 96 S. Ct. 1083, 47 L. Ed. 2d 296 (1976), was insufficient, without more, to establish discriminatory purpose. Moreover, according to the district court, plaintiffs did not establish "even an aggregate of the Zimmer factors, their evidence of tenuous state policy and unresponsiveness of school board members to plaintiffs' particularized interests having failed to persuade the court on those issues." Campbell v. Gadsden County School Board, TCA No. 73-177, slip op. at 8 (N.D.Fla. Dec. 5, 1980).

A review of the record discloses that the plaintiffs in this action did not rely solely on the Zimmer factors5 to establish by circumstantial evidence the presence of discriminatory motivation in the enactment or operation of the voting system. Direct evidence of discriminatory intent in the enactment of the election scheme was presented by plaintiff's expert witness, Dr. Shofner. The district court, however, found that his testimony was "simply not sufficient" to carry plaintiff's burden on the intent issue. Id. at 3. The trial court's finding on this question must be upheld unless it is clearly erroneous. Rogers v. Lodge, --- U.S. ----, ----, 102 S. Ct. 3272, 3279, 73 L. Ed. 2d 1012 (1982); Pullman-Standard v. Swint, --- U.S. ----, ----, 102 S. Ct. 1781, 1788-91, 72 L. Ed. 2d 66, 78-81 (1982).

This is not the first time that a court has addressed the motivation behind the enactment of the present school board election scheme in Florida. In McMillan v. Escambia County, 638 F.2d 1239, 1245-46 (5th Cir. 1981) (Escambia I ), the former Fifth Circuit concluded that it was clear beyond peradventure that the at-large school board primary system had been adopted with an invidious motivation. From 1901 until 1945, the contrivance of the all-white Democratic primary in Florida effectively denied blacks access to the only elections that had substantial meaning because opposition to the Democratic party in the general election was virtually nonexistent. In 1907, the Florida legislature enacted a single-member district plan for school board and county commission primary elections. 1907 Fla. Laws, ch. 5697, § 1. Although the general elections were conducted on an at-large basis, the effect of the all-white Democratic primary was to make the school board "a de facto, if not de jure, single-member district body." Escambia I, 638 F.2d at 1245.

In 1945, following the Supreme Court's decision in Smith v. Allright, 321 U.S. 649, 64 S. Ct. 757, 88 L. Ed. 987 (1944), the Florida Supreme Court outlawed the white primary. Davis v. State ex rel. Cromwell, 156 Fla. 181, 23 So. 2d 85 (1945) (en banc). In the very next legislative session, the Florida legislature enacted statutes requiring both primary and general elections to be conducted at-large. 1947 Fla.Laws, ch. 23726, §§ 7, 9. Dr. Shofner testified in this case, and the former Fifth Circuit concluded in Escambia I, that the change had been made to dilute the growing strength of the black vote. "Looking at the change from single-member districts to at-large districts through Arlington Heights glasses, the conclusion that the change had an invidious purpose is inescapable." Escambia I, 638 F.2d at 1245.

Accordingly, we determine that the district court clearly erred in finding that the 1947 change was made without regard to the effect that it would have in diluting the black vote.

That the at-large election plan was adopted with the motivation of diluting the votes of the minority is only the first prong of the test that must be satisfied. The plaintiff must also demonstrate that the legislation has had the effect of diluting minority votes. Dilution is "generally proven by evidence that a substantial minority is consistently unable to elect candidates of its choice." McMillan v. Escambia County, 638 F.2d 1239, 1248 n.18 (5th Cir. 1981). The plaintiff here has attempted to demonstrate the impact of the at-large system by adducing proof of racially polarized voting patterns and by demonstrating the consistent defeat of black candidates. As in Escambia I, the essence of plaintiff's claim is that because of the racial polarization of voting and the existence of the at-large election system, blacks in Gadsden County, although they comprise a significant portion of the population, are unable to elect members of their race to the school board and "hence, their votes are worth less than those of their white counterparts." Escambia I, 638 F.2d at 1242.

In Rogers v. Lodge, --- U.S. ----, ----, 102 S. Ct. 3272, 3279, 73 L. Ed. 2d 1012 (1982), Justice White, speaking for the Court, observed that "(v)oting along racial lines allows those elected to ignore black interests without fear of political consequences, and without bloc voting the minority candidates would not lose elections solely because of their race." In Gadsden County, black candidates have lost solely because of their race, as the plaintiff's statistical evidence bears out. Blacks comprised 48.5 percent of the registered voters in the county as of October 1978, yet they have been consistently unable to elect candidates of their own race due to the extremely high degree of racial polarization in the voting patterns.

The same statistical tool that was utilized in Escambia I -corroborational analysis-was employed here to demonstrate the degree to which members of a particular race voted for candidates of the same race. Corroborational statistical analysis yields a coefficient factor, R 2, which is a percentage measure of the variation between the percentage of whites registered in each precinct with a percentage of the precinct vote for the white candidates.6 The R 2 factors derived by plaintiff's expert illustrate a higher degree of racial polarization in Gadsden County than the district court found sufficient in Escambia I. Moreover, a comparison of the R 2 factors between the two counties substantiates a wider variation in Escambia County than in Gadsden County. Thus, the evidence of racial polarization is much more compelling in this case.

B. Inability of Blacks to Elect Black Candidates

Since 1972, fourteen blacks have sought county-wide elective office in Gadsden County.7 Of these fourteen, only Harold Henderson, a candidate for the Gadsden County School Board in 1978, was successful. The district court found that Henderson's election cast into doubt plaintiff's statistical evidence as well as the claim that the at-large system itself diluted minority votes. The election of a single black, however, is not conclusive evidence that the votes of the minority are not being diluted. While the lower court was correct in focusing upon the present-day situation in Gadsden County, it attached inordinate significance to Henderson's election. In Zimmer v. McKeithen, 485 F.2d 1297 (5th Cir. 1973) (en banc), aff'd sub nom. East Carroll Parish School Board v. Marshall, 424 U.S. 636, 96 S. Ct. 1083, 47 L. Ed. 2d 296 (1976), the former Fifth Circuit noted the possible divergent explanations for the success of a single black candidate. Such success might be attributable to political support that was motivated by a desire to thwart a successful challenge to the electoral scheme on dilution grounds. Additionally, politicians might find it politically expedient to have a "token" black school board member. Id. at 1307. Plaintiff's expert witness in this case attributed Mr. Henderson's success to characteristics and qualities belonging exclusively to Mr. Henderson that were not transferable to other black candidates in Gadsden County. Indeed, he found it highly improbable that another black candidate would be successful in Gadsden County under the present electoral system.

The district court relied heavily on its finding that the school board members were responsive to the needs of the black community. This responsiveness, appellees contend, precludes any finding of differential impact. Responsiveness or lack thereof, however, goes to proving discriminatory intent in the maintenance of the electoral system-a question not before this court. It has nothing to do with impact. "Whether current office holders are responsive to black needs and campaign for black support is simply irrelevant to that inquiry." Escambia I, 638 F.2d at 1249. In finding that there was no differential impact in Gadsden County, the district court based its decision on its finding of responsiveness to the exclusion of plaintiff's evidence of racially polarized voting patterns and the fact that blacks have been consistently unable to elect a member of their race to the Gadsden County School Board. We hold that the district court's finding was clearly erroneous.

Having found that the at-large school board electoral system in Gadsden County was enacted pursuant to an invidious purpose and that the system has had the effect of diluting minority votes, we reverse and remand this case to the district court for proceedings not inconsistent with this opinion.