This reapportionment case is again before us. On July 12, 1968, Charles Zimmer sued various officials of Louisiana and East Carroll Parish,1 alleging that the apportionment of the police jury (governing body of the parish) and the school board violated the United States Constitution. On May 11, 1971, Stewart Marshall intervened on behalf of himself and all black voters in the parish, arguing that the at-large system approved as a remedy to Zimmer’s complaint violated the rights of black residents of the parish, secured by the Fourteenth and Fifteenth Amendments. On behalf of all black residents in the parish, he sought a declaratory judgment and asked for a single-member district plan.
At this early point in the opinion we must refer, briefly, to the litigative background. The district court adopted a parish at-large election plan. A panel of this Court affirmed. Zimmer v. McKeithen, 1972, 467 F.2d 1381. The Court en banc reversed and remanded on constitutional grounds. 485 F.2d 1297 (1973). The United States Supreme Court affirmed, but on the alternative ground that the district court abused its discretion in approving multi-member districts absent exceptional circumstances. East Carroll Parish Sch. Bd. v. Marshall, 1976, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296. On remand, the district court adopted the defendants’ plan in preference to the plaintiff’s plan. Each plan provides single-member districts. The districts in each plan appear to be drawn so as to ensure the election of white representatives from certain districts and black representatives from the remaining districts, although defendants’ counsel insists that the districts in his plan were not drawn so as to minimize or dilute the voting strength of the blacks in the parish. The court-approved plan of the defendants would probably produce five white representatives and four black repre*929sentatives; the plaintiff’s proposed plan would probably produce five black representatives. Each appears to have been designed consciously to achieve that end, a kind of rough proportioned representation based on registered voting strength. The plaintiff contends that the defendants limited the number of districts in which there was a black majority by drawing jagged lines which gerrymander the districts in the town of Lake Providence, where the blacks are heavily concentrated. The case raises the question, touched on in Taylor v. McKeithen, 5 Cir. 1973, 499 F.2d 893, whether a district court may gerrymander apportionment to effect proportional racial representation.
I.
East Carroll Parish is a small, predominately rural parish in the northeast corner of Louisiana. Its 436 square miles are bounded on the east by the Mississippi River and on the north by Arkansas. In 1970 its population was 12,884. Nearly half that number lived in the parish seat, Lake Providence. At that time, about 55 percent of the registered voters in the parish were whites. A special census was taken in 1976 which revealed that about 60 percent of the parish population were blacks.2 Whites still constituted a slight majority of the registered voters, comprising 51.8 percent of the total. The blacks were concentrated in Lake Providence, where 68 percent of the population were blacks.
This suit was originally filed in the wake of the Supreme Court’s one-man one-vote decisions. In 1968 there were nine members on the parish police jury and eight on the parish school district. These members were elected from districts composed of the traditional wards. Six of the seven wards elected one representative to each board; the third ward, which included the Town of Lake Providence, elected three police jurors and two school board members.3 The population per representative in the various districts ranged from under 400 to nearly 4,000. In December 1968 a consent decree was entered which changed the elections to both bodies to an at-large system. The school board was expanded to nine members, and the traditional wards were used to provide residency requirements. One member of each body was to come from each ward except ward three, which would have three members. At-large voting solved the one-man one-vote problem: all votes had the same mathematical weight.
After the 1970 census results became available, the district court directed the parties to submit new plans. Marshall, the black intervenor, challenged the at-large plan on the ground that it unconstitutionally diluted the votes of the parish’s large black voting minority. Zimmer, the original plaintiff, withdrew from the case.
The district court held a full trial on Marshall’s contentions. It again approved the at-large plan, finding no evidence of racial dilution in the plan, which had been in effect for several years. This Court affirmed. Zimmer v. McKeithen, 5 Cir. 1972, 467 F.2d 1381. On rehearing en banc the Court disagreed with the panel and reversed the district court. Zimmer v. McKeithen, 5 Cir. 1973, 485 F.2d 1297 (en banc) [Zimmer], Judge Gewin wrote, for a closely divided Court, the opinion which has guided this Circuit in later voting dilution cases. See Nevett v. Sides, 5 Cir. 1978, 571 F.2d 209, 216-17. The Court held that Marshall had proved that the at-large system unconstitutionally diluted the value of black votes.
*930The Supreme Court affirmed in a short per curiam opinion. East Carroll Parish School Board v. Marshall, 1976, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296 [Marshall], The Supreme Court did not reach the constitutional issue considered by this Court. Instead, the Court held that the district court abused its discretion in not ordering single-member districts.4 A court-ordered plan is held to equitable standards more strict than those governing legislative plans.5 While legislatures may use multimember districts, courts may order such a remedy only when special circumstances, absent in East Carroll Parish, are shown.
The Court en banc remanded the case to the district court, this time with instructions to approve a single-member district plan. In April 1976, after an abortive attempt to use a special master,6 the district court required the parties to submit single-member district plans. The issue was argued in June 1976. Earl K. Selle, an expert the defendants employed, testified on behalf of the local bodies’ plan; Stanley Hal-pin, counsel for Marshall, defended his alternative plan.
Each plan called for nine single-member districts. Because the population of the parish is concentrated in Lake Providence, each plan divided the town into at least five districts. Although the plans draw different lines for the rural districts, the controversy between the parties centered on the districts in Lake Providence. Marshall objected to the many-sided districts carved from the black areas of the town, and the inclusion of a predominately white suburban area in district five. Selle defended the lines as necessary for the plan’s “fairness”.
The fairness allegedly embodied in the defendants’ plan rests on proportional racial representation. Because the black community was 48.2 percent of the total registered voters in 1976, the defendants drew district lines to guarantee the blacks four seats, and give them a good chance to capture a fifth. The expert’s testimony makes this manifest.
“SELLE: The analysis indicated we have 40% white population and 60% black. *931The analysis of voter registration indicated we have 51.8% white in registration numbers and 48.2% black. Therefore, 60% of the population which is black acquits to the same geological [SIC] structure of 48.2% of the registration of blacks. 60% black population and 48.2% black registration are conversely 40% of your white population developed to 51.8% in your registration.
In the development of our structure, it appeared to me that we would be able to provide very closely the same representation to the blacks as they have actually in the parish, which politically is 48.2%. 48.2% of a nine member board is 4.336 [4.338], I believe, or 4Vs of the seats of a nine member board.
Our plan does provide for four districts, basically six, seven, eight and nine with black majorities respectively 86%, 99%, 64% and 93%.
These four seats, by rule of thumb and practical application and knowledge, would be considered safe districts for the minority group in this parish.
We have District 3, which has a 51% black majority, and if you utilize the same ration with the 51% of population being of the black structure relates back to 40% voting strength within District 3 and 40% is greater than 33. [sic]
This basically, I think, was our fairness test.
If, in fact it is possible to develop a plan which gives each part of the structure or community its rightful justifiable representation — if this is a possibility— then I think it probably should be done.”
Transcript, 34-35.
Selle testified that the odd shapes of some of the districts resulted from both population deviation requirements and racial considerations. Tr. 55. He made no great effort to follow precinct lines, tr. 18, but did, generally, attempt to use natural boundaries, tr. 19. His plan had a maximum population deviation of 6.2 percent; the plaintiff’s plan had a maximum deviation of 9.7 percent, but provided straighter district boundaries within Lake Providence. The defendants’ plan yielded districts with the following racial population compositions:
District 1: 55 percent white, 45 percent black
District 2: 78 percent white, 22 percent black
District 3: 49 percent white, 51 percent black
District 4: 72 percent white, 28 percent black
District 5: 55 percent white, 45 percent black
District 6: 14 percent white, 86 percent black
District 7: 1 percent white, 99 percent black
District 8: 36 percent white, 64 percent black
District 9: 7 percent white, 93 percent black.
Tr. 29-31.
Selle testified that his plan was drawn based on numbers of registered voters. The population of the parish, according to the special 1976 census, was, however, 40 percent white and 60 percent black. Although black voters are now a minority, black residents are now a majority. In oral argument, counsel for each party stated that the evidence would support the conclusion that eventually the registration figures would closely coincide with that population and produce a majority of registered black voters.
The district court adopted the Selle plan, with slight modifications to District 5. The court also awarded the plaintiffs $21,640 as attorneys’ fees, authorized by 42 U.S.C. § 19731(e). This sum represented $40 an hour for the attorneys’ 541 hours of work.
Both sides appealed. Marshall appealed from the plan imposed by the court; the government bodies disputed the amount assessed as attorneys’ fees.
II.
A. Because of the almost infinite number of patterns apportionment might follow in any given geographical area, the trial judge has a wide range of discretion in adopting a plan. The issue here is whether that discretion was abused. In resolving that issue, it is necessary first to decide what standard to apply to this reapportionment plan. Different limits exist to the equitable powers of a district court, imposing its own plan, and the constitutional authority of a state or duly empowered state body to reapportion itself. In Mar*932shall, when this case reached the Supreme Court, the Court based its opinion on equitable considerations without reaching the constitutional questions, citing Connor v. Johnson, 1971, 402 U.S. 690, 91 S.Ct. 1760, 29 L.Ed.2d 268.
More recently, the Supreme Court again stressed the significance of that distinction. In Wise v. Lipscomb, 1978, - U.S. -, 98 S.Ct. 2493, 57 L.Ed.2d 411, the Supreme Court reversed a decision by this Court in the Dallas reapportionment case. The Court held that we improperly considered the plan involved a judicial plan, rather than a legislative plan.
Wise, also a challenge to an at-large voting system, was brought by black and Mexican-American residents of Dallas, Texas. Shortly after certifying a plaintiff class of all black citizens of Dallas, the district court gave “ ‘an opportunity as a legislative body for the City of Dallas to prepare a plan which would be constitutional’ ”. - U.S. at -, 98 S.Ct. 2496. Three days later, the City Council passed a resolution which stated that the Council intended to enact an ordinance establishing eight districts and three at-large seats. Three days later, the City submitted that plan to the district court. After extensive hearings, the court approved the plan which was then passed by the Council. This Court reversed, holding that the plan should be judged as a court-ordered plan. Applying the equitable standards to multi-member districts enunciated in Marshall, we held that the three at-large seats were inappropriate.
The Supreme Court reversed. Justice White announced the decision of the Court in an opinion joined by Justice Stewart. Justice Powell concurred in the judgment in an opinion joined by the Chief Justice and Justices Blackmun and Rehnquist. Justice Marshall dissented, joined by Justices Brennan and Stevens.
Justice White found three important differences between Wise and Marshall. In Wise the district court reviewed the plan as a legislative plan. In Marshall, by contrast, the local bodies submitted plans in response to court orders and “did not purport to reapportion themselves . . ..” - U.S. at -, 98 S.Ct. at 2495. Second, in Marshall federal law prevented the bodies from effectively redistricting. The state law providing police juries and school boards with that power had been disapproved by the Attorney General under § 5 of the Voting Rights Act. Marshall, 424 U.S. at 638 n. 6, 96 S.Ct. 1083. Finally, in Wise the court explicitly gave the Council an opportunity to pass a constitutional apportionment, which it did.
Justice Powell took a different approach. Justice White had relied on the fact that neither the Voting Rights Act nor Texas law prevented the Dallas Council’s reapportionment; the concurring Justices felt that was irrelevant. The process, not the power, controlled.
“The essential point is that the Dallas City Council exercised a legislative judgment, reflecting the policy choices of the elected representatives of the people, rather than the remedial directive of a federal court. As we held in Burns [Burns v. Richardson, 384 U.S. 73, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966)] . . at 85, 86 S.Ct. 1286, at 1293, ‘a State’s freedom of choice to devise substitutes for an apportionment plan found unconstitutional, either as a whole or in part, should not be restricted beyond the clear commands of the Equal Protection Clause.’ This rule of deference to local legislative judgments remains in force even if, as in Burns, our examination of state law suggests that the local body lacks authority to reapportion itself.”
- U.S. at -, 98 S.Ct. at 2501. Justice Powell distinguished Marshall on the ground that the Attorney General’s disapproval of the empowering statute “tainted” the legislative judgments involved. Hence, they could not be afforded the “normal presumption of legitimacy”.
Justice Marshall, in dissent, could not distinguish Marshall from Wise. In neither case, in his view, had the defendants followed valid state procedures for reapportionment; in neither case did the bodies “purport” to be enacting a binding redistricting.
*933Although the deeply split Court in Wise is difficult to follow with confidence, we believe that the facts of this case are indistinguishable from its earlier incarnation in Marshall7
Our main reliance is on the colloquy between the district judge and the attorneys in the case. At the hearing the attorneys were concerned with the possible impact of § 5 of the Voting Rights Act on the defendants’ submissions.
“MR. BRACKEN (attorney for the defendants): The Police Jury and the School Board did adopt resolutions adopting the plan which we refer to as 9-A as the plan to be submitted to the Court. The original resolution also adopted a plan 9-D as an alternate plan. However, at a subsequent joint meeting, plan 9-D was abandoned.
THE COURT: Do you have a copy of your latest resolution?
MR. HALPIN (attorney for the plaintiff): I would like to ask Mr. Bracken to file with the Clerk both resolutions. It is our understanding they have adopted the plan pending this Court’s approval of it.
MR. BRACKEN: Yes.
MR. HALPIN: We will have serious Section 5 problems because when a plan is adopted by a government body, it must be submitted to the Justice Department.
MR. BRACKEN: We submit that all they did was move to submit this plan. That is the only thing adopted.
MR. HALPIN: That is the reason we want to put it in the record as to how it was adopted.
MR. BRACKEN: Your Honor, in this particular proceeding after May 14 when we submitted our particular plan, the resolution was authorizing me to submit to the Court plan 9-A. They did, at that point, authorize us to submit plan 9-D as an alternate.
Upon the submission of an opposing plan by Mr. Halpin, he sent me a copy of an order mailed to you for signature, which you did not sign. That plan called for us to submit any other alternate plan. The bodies were called together and asked whether they wanted me to submit this alternate plan. The second resolution was authorizing me not to submit any alternate plan. That is all the resolution says and that’s all it does.
THE COURT: When the mandate came down from the Supreme Court to the Fifth Circuit and then to this Court, we entered an order that both bodies should submit a plan to the Court. I don’t see where Section 5 cuts any ice in this Court.” [Emphasis added]
Tr. at 4-5. As the underscored portion of this passage makes clear, the defense attorney was careful to avoid suggesting that his clients had adopted the plan. Instead, they merely authorized its submission to the Court. The last statement by the district court also seems to indicate that he saw this as a court-ordered plan. Thus, the first and third factors which Justice White found distinguished Wise from Marshall do not distinguish this case from Marshall.
Furthermore, Justice White’s second distinguishing factor is also absent from this case. The power of the police jury to reapportion a parish is still ruled by the same Louisiana statute, La.Rev.Stat.Ann. §§ 33:1221, 1224 (West Supp.1977). Louisiana is still subject to the pre-clearance provisions of Section 5 of the Voting Rights Act, 42 U.S.C. § 1973c. Thus, the power of the police jury to reapportion is limited by the federal pre-clearance requirement and by the fact that the state statute empowering such changes has been held to violate federal law. Marshall, 424 U.S. at 638 n. 6, 637 n. 2, 96 S.Ct. 1083. The last factor is also important for Justice Powell’s opinion. The factor which he maintained distinguished Wise from Marshall persists in our *934case. By the standards set forth in both prevailing opinions, we think that we have before us a court-ordered plan, to be judged first by equitable standards.
B. The central question in this case, apparently one of first impression, is whether equitable standards permit a district court to approve a plan based on racially proportional representation.
The cases distinguishing between the federal bench’s equitable powers and the constitutional limits on apportionment have held courts to a stricter standard. For example, legislatures may choose multi-member district plans, absent clear violations of the Constitution. Whitcomb v. Chavis, 1971, 403 U.S. 124, 91 S.Ct. 1858, 29 L.Ed.2d 363; White v. Regester, 1973, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314. The judiciary may impose such plans only in the presence of special circumstances. Marshall; Connor v. Finch, 1977, 431 U.S. 407, 415, 97 S.Ct. 1828, 52 L.Ed.2d 465; Chapman v. Meier, 1975, 420 U.S. 1, 95 S.Ct. 751, 42 L.Ed.2d 766. Similarly, legislative plans embodying up to 10 percent deviations in population between districts have been upheld. White v. Regester, 1973, 412 U.S. 755, 93 S.Ct. 2332, 37 L.Ed.2d 314; Gaffney v. Cummings, 1973, 412 U.S. 735, 93 S.Ct. 2321, 37 L.Ed.2d 298. See also Mahan v. Howell, 1972, 410 U.S. 315, 93 S.Ct. 979, 35 L.Ed.2d 320. In Chapman v. Meier, however, (16 percent deviation justified), the Court refused to assume that a 5.95 percent deviation would satisfy the higher standard required of a court-ordered plan. See Connor v. Finch, 431 U.S. at 416-21, 97 S.Ct. 1828. Finally, courts must precisely articulate the interests which justify any deviation from the standards; legislatures need not be candid. Chapman v. Meier, 420 U.S. at 26-27, 95 S.Ct. 751.
The Supreme Court has put tight reins on judicial power to reapportion because of the judiciary’s uneasy position in these cases. The least representative branch of the government must take care when it reforms the most representative branch.
“These high standards reflect the unusual position of federal courts as draftsmen of reapportionment plans. We have repeatedly emphasized that ‘legislative reapportionment is primarily a matter for legislative consideration and determination,’ Reynolds v. Sims, 377 U.S. at 586, 84 S.Ct. 1362 at 1394, 12 L.Ed.2d 506, for a state legislature is the institution that is by far the best situated to identify and then reconcile traditional state policies within the constitutionally mandated framework of substantial population equality. The federal courts by contrast possess no distinctive mandate to compromise sometimes conflicting state apportionment policies in the people’s name. In the wake of a legislature’s failure constitutionally to reconcile these conflicting state and federal goals, however, a federal court is left with the unwelcome obligation of performing in the legislature’s stead, while lacking the political authoritativeness that the legislature can bring to the task. In such circumstances, the court’s task is inevitably an exposed and sensitive one that must be accomplished circumspectly, and in a manner ‘free from any taint of arbitrariness or discrimination’. Roman v. Sincock, 377 U.S. 695, 710, 84 S.Ct. 1449, 12 L.Ed.2d 620.”
Connor v. Finch, 1977, 431 U.S. 407, 414-15, 97 S.Ct. 1828, 1833-1834, 52 L.Ed.2d 465.
The defendants’ plan adopted a straightforward approach to fairness in representation: major groups share representation in the proportions that they share voting strength.8 Although some democra*935cies provide for proportional representation of parties and ethnic groups, it has never been an American tradition. 9 Supreme Court has often held that to make a case *936for unconstitutional districting “it is not enough that the racial group allegedly discriminated against has not had legislative seats in proportion to its voting potential.” White v. Regester, 1973, 412 U.S. 755, 765-66, 93 S.Ct. 2332, 2339, 37 L.Ed.2d 314. See Whitcomb v. Chavis, 1971, 403 U.S. 124, 149, 91 S.Ct. 1858, 29 L.Ed.2d 363; Zimmer, 485 F.2d at 1305. On the other hand, although the constitutionality of a completely racially based apportionment might be suspect, it seems clear that a legislature may take race into account in districting. United Jewish Organizations of Williamsburg v. Carey, 1977, 430 U.S. 144, 97 S.Ct. 996, 51 L.Ed.2d 229. In UJO the Court adopted in the context of race-conscious districting its earlier words about party-conscious districting:
“[C]ourts have [no] constitutional warrant to invalidate a state plan, otherwise within tolerable population limits, because it undertakes, not to minimize or eliminate the political strength of any group or party, but to recognize it and, through districting, provide a rough sort of proportional representation in the legislative halls of the State.”
430 U.S. at 168, 97 S.Ct. at 1011, quoting Gaffney v. Cummings, 1973, 412 U.S. 735, 752, 93 S.Ct. 2321, 37 L.Ed.2d 298.
At this time, we read the decisions of the Supreme Court as admonishing lower federal courts to act cautiously in reapportionments and to leave racially proportional representation to legislative bodies, at least in the absence of some impelling reason to take it into account, for example, where the correction of historic racial discrimination and not merely proper representation is involved.
Our conclusion is reinforced by a consideration of the differences between racially proportionate redistricting and remedial racial goals in other fields. This Court has approved the use of racial goals as remedial devices in employment and education contexts. All other things being equal, it may be assumed that if, for example, 30 percent of a school district’s teaching staff is black, about 30 percent of the teaching staff at any given school should be black. Similarly, absent unlawful discrimination, if 30 percent of the available pool of labor in a plant’s vicinity is black, it could be expected that 30 percent of the plaintiff’s labor force would be black.
Voting is a different matter. The results follow from both the pattern of districts and the voters’ preferences. In employment and education, race may not be a factor in making choices. In voting, while we may deplore the use of race as a factor by voters, its use has not been forbidden. Any such ban would itself face serious constitutional problems. The necessary link between the percentage of those eligible and the percentage actually selected that exists in school cases and employment cases is missing in voting cases. Again, our conclusion is that proportional representation is not appropriate in a court-ordered plan.
This Court has not addressed this precise issue in the past. We have recognized that race may be considered as a factor in determining whether a proposed apportionment is acceptable. “In the process of making a determination [about multimember districts], a court need not be oblivious to the existence and location of minority voting strength.” Zimmer, 485 F.2d at 1308. See Kirksey v. Board of Supervisors of Hinds County, 5 Cir. 1977, 554 F.2d 139, 151 (en banc); Zimmer, 485 F.2d at 1304 n.16. But we have also held that “safe” seats for the minority are not required of a reapportionment plan in a dilution case. United States v. Board of Supervisors of Forrest County, 5 Cir. 1978, 571 F.2d 951, 955.
In a somewhat similár situation in Taylor v. McKeithen, 5 Cir. 1974, 499 F.2d 893, where the reapportionment of four state senatorial districts in New Orleans was before the Court, we disapproved of a court-ordered plan that was designed to ensure the election of a black senator from each of two districts. We observed, “while the Steimel plan gerrymanders and guarantees the election of a black senator from each of districts 2 and 4, this can be accomplished only by diluting black voting power in districts 3 and 5 to the point where the white *937senators from those districts could ignore with impunity the special needs of blacks in those districts.” 499 F.2d at 902. In that case the districts in the plan this Court approved were “compact, contiguous and straight-lined”; taken as a whole, afforded greater access of the minority to the political process than the gerrymandered plan; and the percentage of the black population and registration was rapidly increasing in New Orleans, as it is in East Carroll Parish, leading to the conclusion that some of the districts the trial judge considered “safe” for white candidates were not in fact safe or soon would not be safe. Indeed, one of the weaknesses in racial gerrymandering is that demographic patterns are like shifting sands. See Taylor, 499 F.2d at 499.
We think the cases from this Court are consistent with the middle course we chart today. The district judge must be mindful of the impact of the proposed plans on different racial groups. His duty to avoid both gerrymanders and racial dilution requires that much. The judge must analyze the plan and determine that the probable results are such that minority strength is not diluted. But this legitimate concern with the outcome cannot justify a strict proportionality brought about by manipulation of district lines. If the plan passes the dilution test, as explained in Zimmer, Kirksey, and Nevett v. Sides, 5 Cir. 1978, 571 F.2d 209, race is no longer an important factor. The boundaries should be drawn with an eye to compactness, contiguousness, and the preservation of natural, political, and traditional boundaries; not racially balanced representation. We are not legislatures.
The racially balanced representation sought by the district court’s plan is a plausible solution, but it has no roots in our past. Moreover, it is based on voter registration, not population. The Supreme Court has recognized that reapportionment based on voter registration rather than population may “perpetuate underrepresentation of groups constitutionally entitled to participate in the political process, or perpetuate a ‘ghost of prior malapportionment’ ”. Burns v. Richardson, 1966, 384 U.S. 73, 92-93, 86 S.Ct. 1286, at 1296-1297, 16 L.Ed.2d 376 (footnote and citation omitted). Therefore, reapportionment based on voter registration is allowable only if it produces “a distribution of legislators not substantially different from that which would have resulted from the use of a permissible population basis”. 384 U.S. at 93, 86 S.Ct. at 1297. See Ely v. Klahr, 1971, 403 U.S. 108, 115 n. 7, 91 S.Ct. 1803, 29 L.Ed.2d 352. The district court’s plan, then, is infirm under this standard as well.10 Finally, we point out that not only was the plan racially conscious, but it attempted in effect to assure that the black population majority would remain a voting minority by protecting the white voting majority from erosion.
The plan was devised explicitly to provide V9 black seats, or 44 percent representation compared to 60 percent population. Under City of Richmond v. United States, 1957, 422 U.S. 358, 95 S.Ct. 2296, 45 L.Ed.2d 245, a Section 5 case, it is inferrable that the imposition of an apportionment scheme designed purposefully to limit minority representation to a ratio lower than the ratio of black to white population is unconstitutional.
The lower court did not have before it an acceptable alternative for a “court-ordered plan”. The plaintiff’s proposal contained a maximum population deviation of 9.7 percent which, under Chapman v. Meier, supra, would seem to be unacceptably high, and, in any case, has not been justified by any significant state policy. The smaller deviation evident in the defendants’ plan sug*938gests also that 9.7 percent is unacceptably high, and, if a lower deviation is possible, a desire for proportional representation would not appear, under Chapman, to justify any greater disparity.
We are thus compelled to remand for the formulation of a new plan. This new plan must comply with the traditional standards for “one-man, one-vote” cases, and it must avoid diluting the potential voting strength of the black registration minority, population majority. In this case, the two constitutional duties, assuring each voter the right to an equal vote and avoiding racial discrimination, mesh because the “minority race” actually constitutes the majority of potential voters; the very thrust of “one-man, one-vote” is the prevention of majority dilution through apportionment techniques.
C. We recognize that the district court may have difficulty applying our conclusions to the facts of this case. There is no dispute that the plan adopted by the district court was drawn for the express purpose of providing representation in proportion to each race’s share of the registered vote. The plan offered by the plaintiff, which did not adhere to a strict racial quota, had a roughly equivalent population deviation and considerably more rational boundary lines within the town. The inclusion of the white area north of the lake in district five and the strange boundaries for districts eight and nine resulted from following this scheme.11 On remand the district court should give more weight to the neutral values of rational boundaries, and to districts of equal voting population, and less weight to the political objective of proportional racial representation.
III.
The district court awarded Marshall $21,640 as attorneys’ fees. The police jury and school board appeal. The defendants do not challenge the applicability of the attorneys’ fee statute, 42 U.S.C. § 19731 (e), nor the propriety of an award in this case. Instead, they urge that the award was excessive. They rely on the fact that counsel for Marshall has handled many reapportionment cases and thus was familiar with the legal issues involved. This case has been before our Court three times, once en banc. Marshall and his attorney carried it to the Supreme Court. It has occupied the district court for several years. Regardless of the experience of the counsel, this litigation cannot be termed “routine”. Under the standards set out in Johnson v. Georgia Highway Express Co., 5 Cir. 1974, 488 F.2d 714, the district court was well within its discretion in making this award.
IV.
This case well illustrates this Court’s conclusion in Kirksey:
“Achieving one-man, one-vote political democracy without excluding minorities from political life is a complex task that challenges the best of intellects and requires examining many facets of the community, past, present and future. The problem is not susceptible of simplistic solutions, however seductive they may appear. No mechanistic solution is an alchemistic philosopher’s stone that will turn all the problems of the past and present to future gold.”
554 F.2d at 152. Proportional racial representation, though attractive, is an abuse of the district court’s equitable discretion.
*939The district court decision is REVERSED and REMANDED.
. The named defendants included the Governor of Louisiana, then John McKeithen; the Attorney General; the Secretary of State; tho State Custodian of Voting Machines', the Democratic Committee of East Carroll Parish, its chairman and secretary; the Board of Supervisors of Elections of East Carroll Parish, its president and members; the East Carroll Parish Registrar of Voters; the police jury and school board of the parish, all their members and officials. The case has been litigated by the police jury and the school board.
. The census had two purposes. State road fund money is appropriated in part on the basis of population figures. The state relies on its own figures, but allows parishes to challenge those figures with their own census. The other purpose was to provide information for this lawsuit. Tr. at 25-26.
. A third police juror was added to Ward 3 in 1967 under La.Rev.Stat.Ann. § 33:1223. The juror was elected in 1968. Although the apportionment of parish school boards is dependent on the apportionment of the police jury, the school board had not added a member when this case was first tried. Record, Vol. I at 62. The first reapportionment plan approved by the district court provided for a ninth school board member. Record, Vol. I at 109.
. The Court said: “[The Court of Appeals] seemingly held that multimember districts were unconstitutional, unless their use would afford a minority greater opportunity for political participation, or unless the use of single-member districts would infringe protected rights. . . . [We] now affirm the judgment below, but without approval of the constitutional views expressed by the Court of Appeals. . . . And the Court of Appeals, inexplicably in our view, declined to consider whether the District Court erred under Connor v. Johnson, 402 U.S. 690, 91 S.Ct. 1760, 29 L.Ed.2d 268 (1971), in endorsing a multimember plan, resting its decision instead upon constitutional grounds. We have frequently reaffirmed the rule that when United States district courts are put to the task of fashioning reapportionment plans to supplant concededly invalid state legislation, single-member districts are to be preferred absent unusual circumstances . . . . As the en banc opinion of the Court of Appeals amply demonstrates, no special circumstances here dictate the use of multimember districts. Thus, we hold that in shaping remedial relief the District Court abused its discretion in not initially ordering a single-member reapportionment plan.” 424 U.S. at 638-39, 96 S.Ct. at 1084-1085. (Citations omitted.)
Although the Supreme Court did not rule on the merits of our constitutional holding in Zimmer, we have since reaffirmed that holding. See Blacks United for Lasting Leadership v. City of Shreveport, 5 Cir. 1978, 571 F.2d 248, 251-53.
Four justices expressed the view in Wise v. Lipscomb, 1978, - U.S. -, 98 S.Ct. 2493, 57 L.Ed.2d 411 that the applicability of voting dilution doctrine to municipal governments was unsettled. - U.S. at -, 98 S.Ct. 2493. In this circuit it is clear that the dilution doctrine applies to counties; Zimmer; Kirksey v. Board of Supervisors of Hinds County, 5 Cir. 1977, 554 F.2d 139 (en banc); and cities. Lipscomb v. Wise, 5 Cir. 1977, 551 F.2d 1043; Nevett v. Sides, 5 Cir. 1978, 571 F.2d 209.
. Later, in Connor v. Finch, 1977, 431 U.S. 407, 414, 97 S.Ct. 1828, 1833, 52 L.Ed.2d 465, the Court expressly stated that in discharging the task of reapportioning a state plan, “a court will be held to stricter standards in accomplishing its task than will a state legislature”.
. In response to a motion by Marshall for compliance with the Supreme Court’s mandate, the district court issued guidelines for a special master, then appointed Selle to that position. The court later discovered that Selle had already been employed by the defendant bodies, and revoked his appointment.
. Marshall is not the law of the case on this point because it dealt with an earlier submission by these parties. However, to the extent that this submission resembled the one considered in Marshall, Marshall is a particularly convincing precedent. Our discussion in the text following this deals with the principles by which a court decides whether the plan is court-ordered, and that guides us in concluding that this plan is court-ordered.
. This assumes, of course, that the proportional representation scheme will be successful. That assumption is dubious. The Selle plan used population figures for each of the nine districts and voter registration figures for the entire parish. The plan assumed that in each district the blacks and whites were registered in the same proportion that they were registered in the entire parish. The plan further assumed that the racial groups would vote only for members of their own race. The finding made earlier in this case that East Carroll Parish has a history of racial bloc voting does not necessarily mean that the blocs would be solid if an exceptionally popular and well qualified candidate of one race ran against an exceptionally *935unpopular and unqualified candidate of another race. The plan also required that the same proportion of black and white registered voters actually vote in each election. Each of these assumptions is questionable; combined, they render the predictions of the plan speculative. During oral argument we were informed that elections held under this plan had produced not “4'A” black police jurors, but 3. Even if the Selle plan had produced the racial balance it aimed at in the first elections, changes over time in any of the relevant factors would ruin its scheme. The growing black population in the parish and the increasing registration of black voters suggest that reapportionment based on 1976 figures will not bring about proportionate racial representation in 1980 or 1984.
. An extreme example was the Lebanese system. From its independence, after World War II, until the breakdown of government during its recent civil war, the Lebanese Constitution provided for a unicameral legislature with six Christian representatives for every five Moslem representatives. After the expansion of the legislature to 99 members in 1960, the chamber had to include 30 Maronite Christians, 20 Sunni Moslems, 19 Shi'ite Moslems, 11 Greek Orthodox, 6 Druzes, 6 Greek Catholics, 4 Armenian Orthodox, and 3 members of other minorities. An unwritten convention provided that the President of the Republic would be a Maronite Christian, the premier a Sunni Moslem, and the Speaker of the Chamber a Shi‘ite Moslem. The Cabinet was required to reflect the sectarian balance. 10 Encyclopedia Brittanica 767-68 (1974).
Representation proportionate to a party’s voting strength is more common. The upper house of Japan’s bicameral legislature elects 100 of its 252 members from a national constituency, in order to provide some proportionate representation for political parties. 10 Encyclopedia Brittanica 52 (1974). Similarly, West Germany’s Bundestag is composed of members from districts and members elected at-large from the lander (states). A party must gather at least 5 percent of the vote to be entitled to representation in the Bundestag. 8 Encyclopedia Brittanica 62-63 (1974). In Sweden “each political party receives the same proportion of seats as it receives votes in the general election”. 17 Encyclopedia Brittanica 852 (1974). A party must get either 12 percent of the vote in any one district or 4 percent of the vote throughout the nation to be entitled to any representation. Forty of the 350 members of the Riksdag are allocated directly to the parties in order to achieve the “proper” balance.
Some Anglo-American writers have recognized the advantages of proportional representation. John Stuart Mill wrote:
“In a really equal democracy, every or any section would be represented, not disproportionately, but proportionately. A majority of the electors would always have a majority of the representatives, but a minority of the electors would always have a minority of the representatives. Man for man, they would be as fully represented as the majority. Unless they are, there is not equal government, but a government of inequality and privilege: one part of the people rule over the rest: there is a part whose fair and equal share of influence in the representation is withheld from them, contrary to all just government, but, above all, contrary to the principle of democracy, which professes equality as its very root and foundation.”
J. S. Mill, Considerations on Representative Government 146 (New York 1862). In this country Thomas Hare proposed a system of proportional representation about the same time. T. Hare, Election of Representatives (1865). The Hare system calls for preference voting in multi-member districts. Although it has been tried in some cities and states, it has never been used for congressional elections. Comment, Political Gerrymandering: A Statutory Compactness Standard as an Antidote for Judicial Impotence, 41 U.Chi.L.Rev. 398, 402 n. 21 (1974). See also C. Hoag & G. Hallett, Jr., Proper Representation (2d ed. 1940). Hoag and Hallett support the use of the Hare system, based in large part on the experience of Cincinnati with the plan. Cincinnati, and most of the other jurisdictions in the United States that have tried the system, have since abandoned it. Comment, Political Gerrymandering, 41 U.Chi. L.Rev. 398, 401-04.
This issue has been discussed in the racial context in several articles. See Halpin and Engstrom, Racial Gerrymandering and Southern State Legislative Redistricting: Attorney General Determinations Under the Voting Rights Act, 22 Jour.Pub.Law 37 (1973) (approving proportional representation, but conceding that the “Supreme Court has not indicated any inclination to sustain a racial gerrymandering plan”. Id. at 41); Political Gerrymandering, cited above; Note, Compensatory Racial Reapportionment, 25 Stan.L.Rev. 84 (1972) (supporting proportional representation), Note, Proportional Representation by Race: The Constitutionality of Benign Racial Redistricting, 74 Mich.L.Rev. 820 (1976) (attacking such plans); and Note, United Jewish Organizations v. Carey and the Need to Aggregate Voting Rights, 87 Yale L.J. 571 (1978). See also Dixon, The Court, The People, and “One Man, One Vote” in Reapportionment in the 1970’s, 7 (N. Polsby ed. 1971); Baker, Gerrymandering: Privileged Sanctuary or Next Judicial Target? in Reapportionment in the 1970’s, 121 (N. Polsby ed. 1971).
. Voting age population, rather than voter registration, is an important factor to be considered in assessing the constitutionality of a reapportionment plan. See United Jewish Organizations v. Carey, 1977, 430 U.S. 144, 160-62, 97 S.Ct. 996, 51 L.Ed.2d 229; Beer v. United States, 1976, 425 U.S. 130, 135, 96 S.Ct. 1357, 47 L.Ed.2d 629; Kirksey v. Board of Supervisors, 5 Cir. 1977, 554 F.2d 139, 141, 149-50, cert. denied, 434 U.S. 968, 98 S.Ct. 512, 54 L.Ed.2d 454; Moore v. LeFlore County Board of Election Commissioners, 5 Cir. 1974, 502 F.2d 621, 626.
. Because we have concluded that this plan was adopted to reach an improper goal, we have not had to consider Marshall’s contention that the plan minimized black voting strength. Marshall’s argument is plausible, particularly in light of the extremely high concentration of black voters in districts seven and nine. The most heavily white district, district two, is 78 percent white; the two most heavily black districts are 99 and 93 percent black. The third most heavily concentrated black district, district six, is 86 percent black. While concentrating a race’s population in a few districts makes those districts’ seats “safer”, it also reduces the influence of the group in the remainder of the parish. These concentration figures may be explained by residential patterns in Lake Providence. As we discussed above, the district court on remand should look carefully at the effects of the plans offered to make sure that they are not gerrymanders, minimizing either group’s political power.