dissenting:
Because I believe the majority has created new rules which too narrowly restrict who may be a “minority-preferred candidate” also called “representative of their choice,” and which define “legally significant white bloc voting” in a manner contrary to the emphatic language of the statute, I dissent.
I
As the majority has noted, in this challenge to an at-large voting system under 42 U.S.C. § 1973, complainants must prove (1) that their minority group is sufficiently large and geographically compact to constitute a majority in a single-member district, (2) that their minority group is politically cohesive, and (3) that the white majority votes sufficiently as a bloc to enable it, in the absence of special circumstances, “usually to defeat the minority’s preferred candidate.” Thornburg v. Gingles, 478 U.S. 30, 48-51, 106 S.Ct. 2752, 2765-67, 92 L.Ed.2d 25 (1986). As the majority opinion also recognizes, resolution of this case depends on whether plaintiffs have satisfactorily proven the third prong of this test, legally significant white bloc voting. It is apparent that within the third element are two determinations to be made: first, who can be counted as a “minority-preferred candidate,” and, second, the legal meaning of “usually to defeat” such candidates. The first question is addressed in this section; the second will be considered in Section II.
Unfortunately, neither Gingles nor the statute resolves the precise question of who can be counted as a “preferred candidate.” The most specific formulation of a test on this point has been offered by this court, in Collins IV, 816 F.2d 932, 937 (4th Cir.1987), in which we stated that the support of a successful candidate by a majority of black voters does not necessarily make the candidate “preferred.” In fact, a presumption against such a candidate’s representativeness is invoked if some losing candidate received significantly more minority votes. Collins IV, 816 F.2d at 937. In this case the question is whether two white candidates, Elizabeth Howell, elected to the council in 1974 and 1980, and Claude Staylor, elected to the council in 1974, can be regarded as “preferred candidates” of the minority. As the majority indicates, in 1974, three candidates received over 50 percent of the black vote. They were: Lenious Bond (black, 73.4 percent), Elizabeth Howell1 (white, 58.2 percent), and Claude Staylor, (white, 56.5 percent). Howell and Staylor were elected, and Bond was defeated. In 1980, of the two candidates who received more than 50 percent of the black vote, Evelyn Butts, black, received 92.9 percent and lost, while Howell received 72.9 percent and won. The majority concludes, reasoning from Collins IV, Gingles and two Fifth Circuit decisions, that the district court’s finding that Howell and Staylor were preferred candidates of the minority is an “error of mixed fact and law.”
Although, as the majority recognizes, the legal and factual issues presented are intertwined, the majority result is nonetheless incorrect because it wrongly construes, both legally and factually, the operation of the Collins IV presumption against representativeness. First, implicit in the majority discussion is a legal assumption that in each election there can be only one black-preferred candidate. Collins IV, however, *1245is to the contrary. We specifically referred to the fact that “each such [voting] situation must be reviewed individually to determine whether the elected candidates can be fairly considered as representatives of the minority community.” Collins IV, 816 F.2d at 937. The effect of this language is not to require a finding of “most preferred,” but this is what the majority proposes. The test is whether the elected candidate can be fairly considered as a representative of the minority community and not whether such candidate is the “most preferred.” Moreover, it is worth noting that neither appellants nor the majority seem to object to counting Howell as a black-preferred candidate in 1978, when she received roughly the same support of the black community as she did in 1974 and 1980. The only difference among the elections is that the black candidate, who received the greatest support of the black community in 1978, was elected. In short, Collins IV only creates a presumption, not a rule, that only the most black-preferred candidate can be counted as preferred.
Further, there is no case law to support the idea that preferred means “most preferred.” In fact, the two Fifth Circuit cases cited by the majority stand, in part, for an opposite principle. In Citizens for a Better Gretna v. City of Gretna, 834 F.2d 496, 502 (5th Cir.1987), the court stated that “In a multiple seat election such as Gretna’s aldermanic race, the minority necessarily will have more than one preferred candidate.” The pertinent issue in Gretna was whether a black candidate should be counted as a preferred candidate when he was the only black running for the council, despite the fact that he received a lower percentage of black votes than two white candidates-. The court found that he could be the minority preferred candidate because it was an election for more than one seat and he was the only black candidate. In Campos v. City of Baytown, 840 F.2d 1240, 1245 (5th Cir.1988), the Fifth Circuit held that there may not be a black-preferred candidate when only white candidates sought black or Hispanic votes. Campos carries little weight because in the 1974 and 1980 Norfolk elections, blacks sought election to six of the seven available council seats.
The majority’s citation of Campos and Gretna suggests, albeit implicitly, an additional reason why the majority has construed the legal effect of the Collins IV presumption so decisively against the city. In both Gretna and Campos the Fifth Circuit adopted a rule that the candidate’s race may be taken into account in determining whether he or she is black-preferred. As the court explained in Gretna, “We consider [the black candidate] to be an al-dermanic candidate sponsored by the minority group because he received a significant portion of the black vote, and because he is black.” Gretna, 834 F.2d at 503. Despite the majority’s assertion that it does not decide the issue of whether the race of the candidate is relevant, the harsh way in which the presumption is invoked against Staylor and Howell makes it difficult to conclude that their exclusion as black-preferred candidates is not related to their race. Just as with the majority’s assumption that there can be only one black-preferred candidate, the problem with considering the race of the candidate is that it appears to contradict the law of this circuit. This court, at least implicitly, rejected considering the race of the candidate in Collins IV, 816 F.2d at 937 n. 5, when it stated that: “The question is not, however, so simple as how many blacks versus whites were elected. The [district] court must examine the parties’ studies of voting preferences to determine which were the preferred candidates of the majority and minority communities.”
The effect of the majority’s modification of the Collins TV presumption is plainly revealed in the majority’s discussion of the facts that are said to require the presumption to go unrebutted. The majority first cites the rather conclusory statements of plaintiffs’ witnesses to the effect that they believed Staylor and Howell were not “representatives” of the black community. Considering the fact that “representatives” is a legal term, such, statements should have little weight, particularly when com*1246pared with aggregate voting data indicating majority support for such a candidate.
The remainder of the majority’s evidence redefines the very meaning of electoral representativeness. It finds that Howell and Staylor are not representative because Howell said she does not have an opinion on school busing, that she believed the black community might not have special interests, and because Staylor remarked that he does not admire Dr. Martin Luther King, Jr. The disturbing aspect of this kind of evidence is that it has nothing to do with vote dilution. It only suggests the ideas of certain candidates, and has nothing to do with whether black voters were denied equal access to the election process. Representativeness in a Voting Rights context concerns access, and it does not create a right to the representation of certain ideologies. It is discomforting that the majority seems to be suggesting what minority voters should believe. A consequence of such a rule would be a litmus test for the political beliefs of minority preferred candidates. The entire discussion assumes that in Norfolk there are “proper” black political attitudes, and therefore under the Voting Rights Act some ideas are worth more than others. Ironically, the majority’s position actually debases the votes of those blacks who do not ascribe to the views imputed to their race by the majority. In this context it is worth noting that Evelyn Butts, by the plaintiffs’ own contention, a black-preferred candidate in both 1982 and 1984, testified that she supported continuation of an at-large voting system in Norfolk. According to the majority’s reasoning, it follows that this black woman, who won 87 percent and 72 percent of the votes of the black community in two elections, might not be a black preferred representative.
While focusing on evidence that would redefine vote dilution representativeness in terms of intellectual persuasion, the majority ignores substantial evidence that both Howell and Staylor were in fact the legitimate representatives of the black community. The fact of Ms. Howell’s consistent and substantial black support are overwhelming and uncontradicted. In 1974 and 1980 she was endorsed by the Concerned Citizens of Norfolk (CCN), the city’s most powerful black political group. In 1974, she got more votes than three black candidates. In 1980 she ran better in the black community than one of the two black candidates. She received twice as many black votes as the black president of the Norfolk NAACP. In both elections she received twice as much black support as white support, and she was never supported by more than 39 percent of the whites. The majority’s ruling, as a result, puts her in the absurd position of being overwhelmingly elected three times, but neither black-preferred nor white-preferred. Staylor’s black support, in the 1974 election, is also thoroughly documented. Staylor, the City’s Chief of Police, resigned from a police officer’s organization because it refused to admit blacks. He was endorsed by the CCN.2 In 1974, he defeated three black candidates and seven other white candidates among black voters. He received only about 10 percent more support among white voters than among black voters.
The endorsement by the CCN of these candidates is worth emphasizing. The district court noted that the CCN is the most influential black political group in Norfolk. In 1974, the CCN endorsed Bond, Howell, and Staylor. Three other black candidates were not endorsed by the group. Bond, Howell, and Staylor subsequently received the highest totals of black vote. In 1980, the CCN endorsed Howell and Butts, but did not endorse Banks, the president of the NAACP. Again, the CCN candidates, black and white, outperformed a competing black candidate among black voters. Finally, the majority concludes that an additional reason Staylor and Howell should not be counted as black-preferred is because their vote somehow reflects a forced casting of all ballots, implying that some whites had to get a large number of black votes. This *1247is not supported by the evidence even though these were multi-seat elections. In 1974, Norfolk blacks could have cast all four of their ballots for black candidates, but they chose not to. In 1980, Norfolk blacks could have used the first two of their three ballots to support blacks, but again they chose not to. Similarly, the CCN could have endorsed blacks for six of the seven open seats in 1974 and 1980, but it only endorsed two blacks.
In sum, the exclusion of Staylor and Howell as black-preferred candidates is unwarranted either under the law or under the evidence.3 The majority’s interpretation of the rebuttable presumption of Collins IV assumes that preferred candidate means most preferred and implicitly adopts a rule which considers black candidates more likely to be black-preferred than white candidates. Further, the court, in imposing the presumption, uses an ideological test in order to determine representativeness. This actually dilutes the votes of black citizens, who disagree with the court’s version of black politics, while enhancing the position of a political, not racial group. By emphasizing the ideological test the court has improperly neglected election evidence that Howell and Staylor consistently ran ahead of black candidates among black voters and won the support of the CCN. All of this calls into question the majority’s view of the law and the facts. I feel the district court properly considered the law and I believe its factual findings under this law were not clearly erroneous. I would consider Staylor and Howell as preferred representatives of the black community.
II
Having decided what candidates can be counted as black-preferred, the question becomes whether the white majority votes sufficiently as a bloc to enable it “usually to defeat such candidates.” If Howell and Staylor are considered black-preferred candidates, then the black-community, since 1968, has elected 11 of its 19 preferred candidates. This means that during the same period the black community was represented by one of seven council members (14.3 percent) between 1968 and 1974, two of seven council members (28.6 percent) between 1974 and 1984, and three of seven members (42.9 percent) since 1984. Because black-preferred candidates have won 11 of 19 elections and this is “usual success,” not usual defeat, and because in Norfolk, a city 30 to 35 percent black, these numbers indicate that blacks have been sufficiently and proportionately represented, I believe the district court was correct in determining that plaintiffs had failed to satisfactorily prove the third prong of the Cingles test, the test of legally significant white bloc voting.
Even without the inclusion of Howell and Staylor as black representatives, plaintiffs have failed to show legally significant white bloc voting. It is that part of the majority opinion which addresses this issue that is the most objectionable. It concludes that the test of usual defeat is not a general test, but depends, instead, on blacks’ success in achieving a second seat on the Norfolk council. The special quality of a second seat, although never directly explained by the majority, is that it is the seat which would give plaintiffs proportional representation. The majority concludes that this case is not about electing a single candidate to the council, but is about the black community’s success in choosing a second, that is, proportional representative. The majority does not cite any authority for the proposition that the correct legal test for bloc voting is whether the majority votes sufficiently as a bloc to usually defeat candidates for a proportional seat. There is no such authority. The majority has fashioned a new and sweeping rule of proportionality.
Cingles, unfortunately, does not address the question squarely, but states that a court must determine whether the white *1248majority votes sufficiently as a bloc to usually defeat black-preferred candidates. The rule is expressed only as a general test, effecting a flexible measurement of vote dilution in contests of at-large systems. Moreover, the Gingles court indicated that there is no violation when proportionality is achieved. Therefore one might surmise that under Gingles proportionality and “usual success” are not the same, but represent the upper and lower boundaries of a Section 2 analysis.
The majority’s rule can be distinguished from that stated in Gingles because it is not general or flexible. More important, it is incorrect because it directly contradicts the stated intent of Congress in amending § 2. The principal purpose of the 1982 amendment was to eliminate proof of discriminatory purpose as a prerequisite to establishing a § 2 violation. From the beginning, it was recognized that the elimination of the intent requirement would not confer a right of proportional representation on a minority group. 1982 U.S.Code Cong, and Ad.News 179. When enacted, the statute explicitly stated that “the extent to which members of a protected class have been elected to office in the state or political subdivision is one circumstance which may be considered,” and 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.” 42 U.S.C. § 1973(b).
Although the results test implies that it is necessary, in part, to measure potential violations by candidate success, the denial of a right to proportional representation is also a fundamental, and not an incidental part of the statute. The importance of this limitation was repeated often in the House and Senate discussions on the bill. The Senate Judiciary Committee, which was largely responsible for the form of § 1973(b), made it clear that a denial of equal access was not coequal with the nonexistent right of proportionality. 1982 U.S. Code Cong, and Ad.News 311-320, 363-364, 369. Moreover, not a single senator who voted for the bill argued that a § 2 violation could be shown by a lack of proportional representation. Senator Mitchell stated that the no right to proportionality provision “should dispel any fear about whether this bill will result in proportional representation.” Senator Tsongas noted that the provision should allay any fear that the “ ‘results test’ will have the effect of mandating a system of proportional representation.” Senator Dole, one of the principal authors of the bill, indicated that “No minority group has any constitutional or statutory right to have elected any particular number of their group to political office.” He added that “Equal access is not to be confused with any assurance that minorities must be elected in proportion to the their population.” Senator Kennedy, another major sponsor of the bill, referred to “highly offensive concepts of proportional representation.” Senator Warner stated that the provision was “assurance that the proposed changes in § 2 of the Act would not result in court-ordered establishment of systems of proportional representation by race.” 128 Cong.Rec. 14, 313 (Mitchell); 14,306 (Tsongas); 14, 316-17, 14, 337 (Dole); 14,133 (Kennedy); 14,334 (Warner). All of these statements are the remarks of persons who supported the bill, and they are conclusive evidence that Congress did not intend that a right of proportional representation be established by a court created test.
Although the majority states that its holding is in harmony with Congress’ explicit rejection of proportionality, the nature of the legal test used establishes a right of proportionality. The majority considers the election of one black to the council, even consistently, over a 20 year period, as tokenism. It states that a case to insure the election of one such black “would border on the frivolous.” It focuses, instead, on the “black community’s inability to elect a second representative.” Gingles is interpreted to require scrutiny of white bloc voting “for minority preferred candidates who sought a second seat on the council.” The most pertinent fact, according to the majority, is that between 1968 and 1984 “all of the [black] minority-preferred candidates for a second seat on the council were *1249defeated by candidates preferred by white voters.” These statements indicate that by analyzing the case in such a manner the majority has fashioned a right of proportionality. There is no other way to construe this language than to conclude that the failure to achieve proportionality and the § 2 violation are one and the same. As such, the decision does not merely creep up to the notion of proportionality, it embraces it.
The effect of such a rule extends far beyond this case. The majority decision guarantees proportionality in every jurisdiction in this circuit, because it concludes that the usual failure of a minority candidate who is running for a proportional seat is illegal vote dilution. Any at-large system, which does not usually produce proportional representation, is now illegal. Such a holding ignores the clear intent of Congress, and fails to offer any meaningful judicial compromise of the tension created by the coexistence in § 2 of the results test and the no right to proportionality provision. Here, all doubt is resolved in favor of proportionality.
A proper test, which takes into account the meaning of the entire statute, would inquire as to whether black-preferred candidates are normally, or usually defeated. Such a rule does not measure a violation by a right of proportionality, which does not exist. By taking into account the historic success or failure of all black candidates, it would also provide a more accurate portrayal of vote dilution. It would recognize that the election of a single black is not unimportant, because in an at-large system the majority always has the strength to elect all of its candidates. Such a rule would also take account of those situations, as in Norfolk, where, because of its size, the black community is likely to seek more than one seat. It would consider that when there are more defeats of black-preferred candidates, the evidence is more compelling that the system is operating to dilute black votes. See Gingles, 478 U.S. at 56, 106 S.Ct. at 2769 (listing the kinds of factors that should be considered in a white bloc voting analysis). The effect of such a rule would be to give real meaning to all of the statutory language and to Congressional intent by accommodating both the results test and the denial of a right of proportionality.
Under such a test, between 1968 and 1986, 7 of the 13 black-supported candidates were elected to office. Therefore blacks have not normally or usually been defeated. If Howell and Staylor are considered as black-preferred candidates, then such candidates have won 11 of 19 elections during the same period. These successes are enough to defeat plaintiffs claim of “legally significant white bloc voting.”
Ill
Although the most far reaching aspects of the majority’s opinion are discussed in Sections I and II, the majority’s treatment of several factors as “special circumstances” to explain black success, deserve comment because they indicate a more general problem of the manner in which the majority’s treatment of factual issues in this case tends to limit the deference due the district court findings.
The majority discounts the success of Reverend Foster in 1984 by ascribing it to the white Norfolk Mayor’s endorsement of Foster, and indicates that the Mayor was motivated by an attempt to moot this lawsuit. As this court held in Collins IV, 816 F.2d at 938, however, “the Mayor’s intent, or whether he or other leaders conspired, is not dispositive” of the legal issue of whether the election was a “special circumstance.” Instead, “A proper inquiry must examine the result of the Mayor’s conduct and statement. If voting patterns show unusual white support for the black candidate in 1984, the legal significance of his success should be diminished.” A finding of unusual white support should be disturbed only if clearly erroneous. Collins IV, 816 F.2d at 938. The majority holds that the district court committed legal error because it measured the white support for Foster in 1984 against white support for all other black candidates, and not just against black candidates contending for the “second seat.”
*1250First, this analysis is incorrect on the ground, already stated, that it establishes a right to proportional representation. Foster’s success should be compared against the white vote for all blacks, not just those running for the proportional seat. This evidence indicates that Foster’s level of white support was not unusual at all. Green (1978, 1982, 1986) and Jordan (1968) are blacks who received greater white support than Foster. Moreover, Foster’s success was statistically insignificant when compared to the performance of black-preferred candidates in four other elections. Foster’s white support was either less than or equal to white support for other black candidates in eight of thirteen elections. The district court’s finding that there was no “special circumstance” in Foster’s election is not clearly erroneous.
More seriously, the majority’s conclusion is incorrect even on its own terms. In the same 1984 election, Evelyn Butts, a black seeking election to the same “second seat,” received 24.2 percent of the white vote. This total is statistically insignificant from Foster’s 26.6 percent figure. The majority’s effort to explain this by stating that unusual support for Butts was also evidence of unusual white support for Foster, the elected candidate, is unfathomable. First, the Mayor and his supporters’ acted for Foster, not Butts. Their action in behalf of Foster is the sole basis for this special circumstance inquiry. Thus, the question is whether the mayor’s support produced unusual support for Foster, not Butts. The implication of Butts’ relative success among white voters is plainly that the Mayor’s actions failed to produce unusual support for Foster because Butts performed just as well among whites as Foster. White votes for Butts did not help Foster to get elected, so this did not produce a so-called “special circumstance.” Under any test, the district court was not clearly erroneous in finding Foster’s election was not the result of “special circumstance.” Indeed, the nature of the dispute over what the evidence reveals emphasizes why fact finding is entrusted to the trial judge.
Second, the majority holds that the district court’s finding that incumbency was not a special circumstance was clearly erroneous. The majority’s test seems to be that the electoral success of incumbents and the defeat of nonincumbents indicates that incumbency rises to the level of a special circumstance. Such a rule would make practically every American election a “special circumstance.” For “special circumstance” to have any meaning one must produce specific evidence that incumbency, in the case of individual candidates, produced real benefits. The evidence here is that it did not. The majority points out that “with two exceptions” “no black candidate who was not an incumbent was elected.” The problem with this conclusion is that between 1968 and 1988 there were only six black-preferred candidates, who ran as unelected non-incumbents, and two of them were elected. The majority seems to reason if perhaps four, five, or all six non-incumbents were elected, then incumbency could not be considered a special circumstance. Given the nature of the election process, the expectation is unreasonable, particularly because the pertinent question is whether non-incumbency itself is an unusual handicap in Norfolk. The majority also states that no incumbent black has ever lost a reelection campaign. The test, however, is not merely who wins and loses, but whether incumbency helps to explain such wins and losses. The statistics indicate that the two blacks who were reelected as incumbents, Jordan and Green, did not benefit from incumbency. For example, Jordan’s white support declined in both 1972 and 1976 from his first election in 1968. He was nonetheless reelected. Similarly, Green’s white support declined from 1978 levels in the 1982 and 1986 election. However, he was reelected. Green even lost black support during the same period. Declining support, particularly among whites, is not a measure of the benefits of incumbency. Instead, it raises a reasonable inference that these men were reelected based on a core constituency developed in their first election, not in the period of incumbency. The district court’s finding on this point was not clearly erroneous.
*1251The judgment of the district court was correct, and it should be affirmed. The majority opinion, in several respects, creates new law which I believe conflicts with the law of this circuit, the rule of Gingles, and the intent of Congress. For these reasons and the reasons stated above, I respectfully dissent.
. Mrs. Howell ran ahead of three black candidates in the black community and she received a higher percent of the black votes than white votes.
. Plaintiff Collins testified that Staylor was one of three candidates of choice among the blacks in 1974.
. Reverend Green testifying about Staylor’s resignation from the all white police officers organization stated: “The black community thought it was a brave statement, and when he ran for council, they voted for him overwhelmingly. He received the largest vote any candidate has ever received for City Council in the history of the election of the members of City Council.”