concurring and dissenting in part:
I. Liability
A determination by a federal court that elected officials have intentionally discriminated against some of their constituents is a matter of no little moment. While I join the liability portion of Judge Schroeder’s opinion without reservation, I write briefly to explain, for the benefit of those not conversant with the esotérica of federal discrimination law, what today’s ruling means — and what it does not.
First the good news. Nothing in the majority opinion, or in the district court’s findings which we review and approve today, suggests that the County supervisors who adopted the 1981 reapportionment — all of whom are still in office — harbored any ethnic or racial animus toward the Los Angeles Hispanic community. In other words, there is no indication that what the district court found to be intentional discrimination was based on any dislike, mistrust, hatred or bigotry against Hispanics or any other minority group. Indeed, the district court seems to have found to the contrary. See Garza v. County of Los Angeles, Nos. 88-5143 & 88-5435, at 7 (C.D.Cal. June 4, 1990) (“The Court believes that had the Board found it possible to protect their incumbencies while increasing Hispanic voting strength, they would have acted to satisfy both objectives.”).1
Which brings us to what this case does stand for. When the dust has settled and local passions have cooled, this case will be remembered for its lucid demonstration that elected officials engaged in the single-minded pursuit of incumbency can run roughshod over the rights of protected minorities. The careful findings of the district court graphically document the pattern — a continuing practice of splitting the Hispanic core into two or more districts to prevent the emergence of a strong Hispanic challenger who might provide meaningful competition to the incumbent supervisors. The record is littered with telltale signs that reapportionments going back at least as far as 1959 were motivated, to no small degree, by the desire to assure that no supervisorial district would include too much of the burgeoning Hispanic population.
But the record here illustrates a more general proposition: Protecting incumbency and safeguarding the voting rights of minorities are purposes often at war with each other. Ethnic and racial communities are natural breeding grounds for political challengers; incumbents greet the emergence of such power bases in their districts with all the hospitality corporate managers show hostile takeover bids. What happened here — the systematic splitting of the ethnic community into different districts— is the obvious, time-honored and most effective way of averting a potential challenge. Incumbency carries with it many other subtle and not-so-subtle advantages, see Chemerinsky, Protecting the Democratic Process: Voter Standing to Challenge Abuses of Incumbency, 49 Ohio St. L.J. 773, 774-81 (1988), and incumbents who take advantage of their status so as to assure themselves a secure seat at the expense of emerging minority candidates may well be violating the Voting Rights Act. Today’s case barely opens the door to our understanding of the potential relationship between the preservation of incumbency and invidious discrimination, but it sure*779ly gives weight to the Seventh Circuit’s observation that “many devices employed to preserve incumbencies are necessarily racially discriminatory.” Ketchum v. Byrne, 740 F.2d 1398, 1408 (7th Cir.1984), cert. denied, 471 U.S. 1135, 105 S.Ct. 2673, 86 L.Ed.2d 692 (1985).
The Supreme Court in Davis v. Bandemer, 478 U.S. 109, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986), left open whether and under what circumstances political gerrymandering may amount to a violation of the Voting Rights Act. Id. at 118 n. 8, 106 S.Ct. at 2803 n. 8. The record before us strongly suggests that political gerrymandering tends to strengthen the grip of incumbents at the expense of emerging minority communities. Where, as here, the record shows that ethnic or racial communities were split to assure a safe seat for an incumbent, there is a strong inference — indeed a presumption — that this was a result of intentional discrimination, even absent the type of smoking gun evidence uncovered by these plaintiffs. State and local officials nationwide might well take this lesson to heart as they go about the task of decennial redistricting.
II. The Remedy
While I enthusiastically join the majority as to liability, I have two points of disagreement as to the remedy. The first is really just a quibble: I agree with the majority that the County’s proposed plan was not entitled to any deference. The Los Angeles County Charter requires at least four supervisors to pass a reapportionment plan. Los Angeles County Charter Art. 2, § 7. Since two of the five supervisors opposed the plan proposed by the County, see maj. op. at 776, it is obvious that the “proposal was not an act of legislation; rather, it was a suggestion by some members of the Board,” id., not entitled to the special deference afforded apportionment plans that are the legislative act of the apportioning body.
The majority’s alternative reason for upholding the district court’s rejection of the plan, contained in the last sentence of part II.D of the opinion, is therefore dicta, and dicta about which I harbor some doubt. It is not at all clear to me that, had the Board of Supervisors adopted the apportionment plan proposed by the County, the reasons relied on by the district court for rejecting the plan would be sufficient. Certainly the issue is far more difficult than the majority’s casual reference acknowledges. I would prefer to see a more detailed discussion of the issue before adopting the majority’s conclusion as the law of the circuit, but a more extensive discussion is inappropriate, as it’s all dicta anyhow. The more prudent course would be' to reserve the issue for a day when it is squarely presented to us.
My second disagreement is more substantive; I cannot agree with the majority’s conclusion, contained in part II.C of the opinion, that the district court’s reapportionment plan complies with the one person one vote principle announced by the Supreme Court in Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964). While the majority may ultimately be vindicated, its conclusion is hard to square with what the Supreme Court has said on this issue up to now.
A. Before plumbing the doctrinal waters in this murky area of constitutional law, it is worth stating exactly what the County is complaining about. In drawing the remedial plan in this case, the district court adhered closely to state law which calls for supervisorial districts that are equal in population. In doing so, the court wound up with two districts where the numbers of voting age citizens are markedly lower than those in the three other districts.2 The disparity is particularly *780great between Districts 1 and 3. District 1 has 707,651 eligible voters while District 3 has 1,098,663, a difference of 391,012, about 55% of the eligible voters in District 1. Since it takes a majority in each district to elect a supervisor, this means that the supervisor from District 1 can be elected on the basis of 353,826 votes (less than the difference between the two districts), while the supervisor from District 3 requires at least 549,332 votes. Put another way, a vote cast in District 1 counts for almost twice as much as a vote cast in District 3.
B. Does a districting plan that gives different voting power to voters in different parts of the county impair the one person one vote principle even though raw population figures are roughly equal? It certainly seems to conflict with what the Supreme Court has said repeatedly. For example, in Reynolds, the Court stated: “Weighting the votes of citizens differently, by any method or means, merely because of where they happen to reside, hardly seems justifiable.” 377 U.S. at 563, 84 S.Ct. at 1382. The Court also stated: “With respect to the allocation of legislative representation, all voters, as citizens of a State, stand in the same relation regardless of where they live,” id. at 565, 84 S.Ct. at 1383; and “Simply stated, an individual’s right to vote for state legislators is unconstitutionally impaired when its weight is in a substantial fashion diluted when compared with votes of citizens living in other parts of the State,” id. at 568, 84 S.Ct. at 1385;3 and “the basic principle of representative government remains, and must remain, unchanged — the weight of a citizen’s vote cannot be made to depend on where he lives,” id. at 567, 84 S.Ct. at 1384.
Almost identical language appears in numerous cases both before Reynolds, see, e.g., Wesberry v. Sanders, 376 U.S. 1, 8, 84 S.Ct. 526, 530, 11 L.Ed.2d 481 (1964) (“To say that a vote is worth more in one district than in another would not only run counter to our fundamental ideas of democratic government, it would cast aside the principle of a House of Representatives elected ‘by the People.’ ”); Gray v. Sanders, 372 U.S. 368, 379, 83 S.Ct. 801, 808, 9 L.Ed.2d 821 (1963) (“Once the geographical unit for which a representative is to be chosen is designated, all who participate in the election are to have an equal vote — whatever their race, whatever their sex, whatever their occupation, whatever their income, and wherever their home may be in that geographical unit.”4); and after, see, e.g., Hadley v. Junior College Dist., 397 U.S. 50, 56, 90 S.Ct. 791, 795, 25 L.Ed.2d 45 (1970) (“[WJhen members of an elected body are chosen from separate districts, each district must be established on a basis that will insure, as far as is practicable, that equal numbers of voters can vote for proportionally equal numbers of officials.”); Chapman v. Meier, 420 U.S. 1, 24, 95 S.Ct. 751, 764, 42 L.Ed.2d 766 (1975) (“All citizens are affected when an apportionment plan provides disproportionate voting strength, and citizens in districts that are underrepresented lose something even if they do not belong to a specific minority group.”); Lockport v. Citizens for Community Action, 430 U.S. 259, 265, 97 S.Ct. 1047, 1052, 51 L.Ed.2d 313 (1977) (“[I]n voting for their legislators, all citizens have an equal interest in representative democracy, and ... the concept of equal protection therefore requires that their votes be given equal weight.”).
The1 Court adhered to the same formulation as recently as two Terms ago: “In calculating the deviation among districts, the relevant inquiry is whether ‘the vote of any citizen is approximately equal in *781weight to that of any other citizen.’ ” Board of Estimate v. Morris, 489 U.S. 688, 109 S.Ct. 1433, 1441, 103 L.Ed.2d 717 (1989) (quoting Reynolds, 377 U.S. at 579, 84 S.Ct. at 1390).
Despite these seemingly clear and repeated pronouncements by the Supreme Court, the majority’s position is not without support, as the Court has also said things suggesting that equality of population is the guiding principle. See, e.g., Reynolds, 377 U.S. at 568, 84 S.Ct. at 1385 (“We hold that, as a basic constitutional standard, the Equal Protection Clause requires that the seats in both houses of a bicameral state legislature must be apportioned on a population basis.”); Mahan v. Howell, 410 U.S. 315, 321, 93 S.Ct. 979, 983, 35 L.Ed.2d 320 (1973) (“[T]he basic constitutional principle [is] equality of population among the districts.”); Kirkpatrick v. Preisler, 394 U.S. 526, 530, 89 S.Ct. 1225, 1228, 22 L.Ed.2d 519 (1969) (“ ‘[E]qual representation for equal numbers of people [is] the fundamental goal for the House of Representatives.’ ” (quoting Wesberry, 376 U.S. at 18, 84 S.Ct. at 535)).
In most cases, of course, the distinction between the two formulations makes no substantive difference: Absent significant demographic variations in the proportion of voting age citizens to total population, apportionment by population will assure equality of voting strength and vice versa. Here, however, we do have a demographic abnormality, and the selection of an apportionment base does make a material difference: Apportionment by population can result in unequally weighted votes, while assuring equality in voting power might well call for districts of unequal population.
How does one choose between these two apparently conflicting principles? It seems to me that reliance on verbal formulations is not enough; we must try to distill the theory underlying the principle of one person one vote and, on the basis of that theory, select the philosophy embodied in the fourteenth amendment. Coming up with the correct theory is made no easier by the fact that the Court has been less than consistent in its choice of language and that, as Justice Harlan pointed out in his Reynolds dissent, “both the language and history of the controlling provisions of the Constitution [have been] wholly ignored” by the Court, 377 U.S. at 591, 84 S.Ct. at 1397 (Harlan, J., dissenting), making it impossible, to rely on the Constitution for any meaningful guidance. Still we must try.
C. While apportionment by population and apportionment by number of eligible electors normally yield precisely the same result, they are based on radically different premises and serve materially different purposes. Apportionment by raw population embodies the principle of equal representation; it assures that all persons living within a district — whether eligible to vote or not — have roughly equal representation in the governing body.5 A principle of equal representation serves important purposes: It assures that constituents have more or less equal access to their elected officials, by assuring that no official has a disproportionately large number of constituents to satisfy. Also, assuming that elected officials are able to obtain benefits for their districts in proportion to their share of the total membership of the governing body, it assures that constituents are not afforded unequal government services depending on the size of the population in their districts.
Apportionment by proportion of eligible voters serves the principle of electoral equality. This principle recognizes that electors — persons eligible to vote — are the ones who hold the ultimate political power in our democracy. This is an important power reserved only to certain members of society; states are not required to bestow it upon aliens, transients, short-term residents, persons convicted of crime, or those considered too young. See J. Nowak, R. *782Rotunda & J.N. Young, Constitutional Law § 14.31, at 722-23 (3d ed. 1986).
The principle of electoral equality assures that, regardless of the size of the whole body of constituents, political power, as defined by the number of those eligible to vote, is equalized as between districts holding the same number of representatives. It also assures that those eligible to vote do not suffer dilution of that important right by having their vote given less weight than that of electors in another location. Under this paradigm, the fourteenth amendment protects a right belonging to the individual elector and the key question is whether the votes of some electors are materially undercounted because of the manner in which districts are apportioned.
It is very difficult, in my view, to read the Supreme Court’s pronouncements in this area without concluding that what lies at the core of one person one vote is the principle of electoral equality, not that of equality of representation. To begin with, the name by which the Court has consistently identified this constitutional right— one person one vote — is an important clue that the Court’s primary concern is with equalizing the voting power of electors, making sure that each voter gets one vote — not two, five or ten, Reynolds, 377 U.S. at 562, 84 S.Ct. at 1381; or one-half.
But we need not rely on inferences from what is essentially an aphorism, for the Court has told us exactly and repeatedly what interest this principle serves. In its most recent pronouncement in the area, the Court stated: “The personal right to vote is a value in itself, and a citizen is, without more and without mathematically calculating his power to determine the outcome of an election, shortchanged if he may vote for only one representative when citizens in a neighboring district, of equal population, vote for two; or to put it another way, if he may vote for one representative and the voters in another district half the size also elect one representative.” Morris, 109 S.Ct. at 1440 (emphasis added).
References to the personal nature of the right to vote as the bedrock on which the one person one vote principle is founded appear in the case law with monotonous regularity. Thus, in Hadley v. Junior College District, the Court stated: “[T]he Fourteenth Amendment requires that the trustees of this junior college district be apportioned in a manner that does not deprive any voter of his right to have his own vote given as much weight, as far as is practicable, as that of any other voter in the junior college district.” 397 U.S. at 52, 90 S.Ct. at 792. The Court further explained: “[A] qualified voter has a constitutional right to vote in elections without having his vote wrongfully denied, debased, or diluted,” id. (footnote omitted); and “This Court has consistently held in a long series of cases, that in situations involving elections, the States are required to insure that each person’s vote counts as much, insofar as it is practicable, as any other person’s,” id. at 54, 90 S.Ct. at 794 (footnote omitted); and “once a State has decided to use the process of popular election and ‘once the class of voters is chosen and their qualifications specified, we see no constitutional way by which equality of voting power may be evaded,’ ” id. at 59, 90 S.Ct. at 797 (quoting Gray v. Sanders, 372 U.S. at 381, 83 S.Ct. at 809).
Reynolds itself brims over with concern about the rights of citizens to cast equally weighted votes: “[T]he judicial focus must be concentrated upon ascertaining whether there has been any discrimination against certain of the State’s citizens which constitutes an impermissible impairment of their constitutionally protected right to vote.” 377 U.S. at 561, 84 S.Ct. at 1381. Again: “Full and effective participation by all citizens in state government requires, therefore, that each citizen have an equally effective voice in the election of members of his state legislature.” Id. at 565, 84 S.Ct. at 1383.6 And yet again: “And the right of suffrage can be denied by a debasement or *783dilution of the weight of a citizen’s vote just as effectively as by wholly prohibiting the free exercise of the franchise.” Id. at 555, 84 S.Ct. at 1378. Reynolds went so far as to suggest that “[t]o the extent that a citizen’s right to vote is debased, he is that much less a citizen.” Id. at 567, 84 S.Ct. at 1384.
While the Court has repeatedly expressed its concern with equalizing the voting power of citizens as an ultimate constitutional imperative — akin to protecting freedom of speech or freedom of religion— its various statements in support of the principle of equal representation have been far more conditional. Indeed, a careful reading of the Court’s opinions suggests that equalizing total population is viewed not as an end in itself, but as a means of achieving electoral equality. Thus, the Court stated in Reynolds: “[T]he overriding objective must be substantial equality of population among the various districts, so that the vote of any citizen is approximately equal in weight to that of any other citizen in the State.” Id. at 579, 84 S.Ct. at 1390 (emphasis added). This language has been quoted in numerous subsequent cases. See Gaffney, 412 U.S. at 744, 93 S.Ct. at 2327; Mahan v. Howell, 410 U.S. 315, 322, 93 S.Ct. 979, 984, 35 L.Ed.2d 320 (1973); Burns, 384 U.S. at 91 n. 20, 86 S.Ct. at 1296 n. 20. In Connor v. Finch, 431 U.S. 407, 416, 97 S.Ct. 1828, 1834, 52 L.Ed.2d 465 (1977), the Court stated the proposition as follows: “The Equal Protection Clause requires that legislative districts be of nearly equal population, so that each person’s vote may be given equal weight in the election of representatives.” (emphasis added).7
Particularly indicative of the subservience of the representational principle to the principle of electoral equality is Gaffney v. Cummings, 412 U.S. 735, 93 S.Ct. 2321, 37 L.Ed.2d 298 (1973), on which the majority mistakenly relies. Gaffney deals with the question of how much variation in population is permissible in effectuating the one person one vote principle of Reynolds. The Supreme Court held that absolute mathematical precision is not necessary. Total population, the Court pointed out, is only a proxy for equalizing the voting strength of eligible voters. But, the Court noted, it is not a perfect proxy; voters might not be distributed homogeneously throughout the population, for example. Therefore, “it makes little sense to conclude from relatively minor ‘census population’ variations among legislative districts that any person’s vote is being substantially diluted.” Gaffney, 412 U.S. at 745-46, 93 S.Ct. at 2327-28. The Court continued:
What is more, it must be recognized that total population, even if absolutely accurate as to each district when counted, is nevertheless not a talismanic measure of the weight of a person’s vote under a later adopted reapportionment plan. The United States census is more of an event than a process. It measures population at only a single instant in time. District populations are constantly changing, often at different rates in either direction, up or down. Substantial differentials in population growth rates are striking and well-known phenomena. So, too, if it is the weight of a person’s vote that matters, total population — even if stable and accurately taken — may not actually reflect that body of voters whose votes must be counted and weighed for the purposes of reapportionment, because “census persons” are not voters.
Id. at 746, 93 S.Ct. at 2328 (emphasis added, footnotes omitted).
Finally, there is the teaching of Burns v. Richardson, 384 U.S. 73, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1966), which the majority dismisses far too lightly. Because it is the only Supreme Court case applying the one person one vote principle in a situation *784where there were large numbers of residents not eligible to vote — it being the only case where there was a divergence between the representational principle and the principle of electoral equality — the case deserves a more careful examination. While Bums does not, by its terms, purport to require that apportionments equalize the number of qualified electors in each district, the logic of the case strongly suggests that this must be so. As noted earlier, in a situation such as ours — as that in Bums — one or the other of the principles must give way. If the ultimate objective were to serve the representational principle, that is to equalize populations, Bums would be inexplicable, as it approved deviations from strict population equality that were wildly in excess of what a strict application of that principle would permit.8
Bums can only be explained as an application of the principle of electoral equality; the Court approved the departure from strict population figures because raw population did not provide an accurate measure of whether the voting strength of each citizen was equal. Thus, while Bums spoke in permissive terms, its logic is far more categorical.
The only other way to explain the result in Bums is to assume that there is no principle at all at play here, that one person one vote is really nothing more than a judicial squinting of the eye, a rough-and-ready determination whether the apportionment scheme complies with some standard of proportionality the reviewing court happens to find acceptable. I am reluctant to ascribe such fluidity to a constitutional principle that the Supreme Court has told us embodies “fundamental ideas of democratic government,” Wesberry, 376 U.S. at 8, 84 S.Ct. at 530.9
When considered against the Supreme Court’s repeated pronouncements that the right being protected by the one person one vote principle is personal and limited to citizens, the various arguments raised by the majority do not carry the day. Thus, the Court’s passing reference in Kirkpatrick v. Preisler, 394 U.S. 526, 531, 89 S.Ct. 1225, 1229, 22 L.Ed.2d 519 (1969), to “preventing] debasement of voting power and diminution of access to elected representatives” suggests only that the Court did not consider the possibility that the twin goals might diverge in some cases. As Kirkpatrick contains no discussion of the issue, it provides no clue as to which principle has primacy where there is a conflict between the two.
Similarly unpersuasive is the majority’s citation of cases that hold that aliens and the young enjoy many constitutional rights on the same basis as citizens. Maj. op. at 775. One right aliens and children do not enjoy is the right to vote. Insofar as the Court views its one person one vote jurisprudence as protecting the right to vote enjoyed only by citizens, see pp. 780-781 supra, it’s entirely beside the point what other rights noncitizens may enjoy. If, as I suggest, one person one vote protects a right uniquely held by citizens, it would be a dilution of that right to allow noncitizens to share therein.10
*785Finally, I understand my colleagues to be suggesting that, as a matter of policy, the principle of equal representation is far wiser than the principle of electoral equality. Were I free to disregard the explicit and repeated statements of the Supreme Court, I might well find this argument persuasive. But I am not free to ignore what I regard as binding direction from the Supreme Court, so my own policy views on this matter make no difference.
All that having been said, I must acknowledge that my colleagues may ultimately have the better of the argument. We are each attempting to divine from language used by the Supreme Court in the past what the Court would say about an issue it has not explicitly addressed. While much of the language and some of the rationale of the Supreme Court’s decisions clearly support my view, other language, as well as tradition, supports my colleagues. Were the Supreme Court to take up the issue, I would not be surprised to see it limit or abandon the principle of electoral equality in favor of a principle of representational equality. But the implications of that decision must be considered by those who have the power to make such choices, not by us. My colleagues may well be looking into the future, but controlling guidance comes from the past.
D. Having concluded that it is the principle of electoral equality that lies at the heart of one person one vote, we must address whether the district court’s plan nevertheless falls within acceptable limits. While the Supreme Court has not been completely consistent in its methodology, usually it creates hypothetical ideal districts (i.e., districts that contain precisely the same number of people) and then determines, in percentage terms, the degree of deviation between each of the actual districts and the ideal one. The maximum deviation is calculated by adding the percentage points that the largest district is above the ideal, to the percentage points the smallest is below. See, e.g., Brown v. Thomson, 462 U.S. at 839, 103 S.Ct. at 2694, Mahan, 410 U.S. at 319, 93 S.Ct. at 982. While the Court has always used raw population figures, not electors, there seems to be no reason to apply a different methodology when comparing numbers of electors.
Here, a hypothetical ideal district would contain 979,594 electors. See note 2 supra. Compared to this ideal district, the districts under the plan adopted below deviate as set forth in the following table:
District # Electors Raw Deviation % Deviation
707,651 -271,943 -28% H
922,180 -57,414 -6% tO
1,098,663 +119,069 +12% CO
1,081,089 +101,495 +10%
1,088,388 +108,794 +11% C71
As this table demonstrates, the districts in the court-ordered plan contain very significant deviations from the ideal district. As expected, the greatest spread is between Districts 1 and 3, and it amounts to 40%. Equally significant are the individual deviations. Only one district, number 2, has a number of electors close to the norm, i.e., a deviation within single digits. Three of the districts have deviations between 10% and 20% and one district has a deviation nearly three times that amount — 28%.
If I am right that it is qualified electors, not raw population figures, that count, these deviations fall far outside the acceptable range. The Supreme Court’s cases in this area have defined three ranges of deviation that bear on the constitutionality of the plan. A maximum deviation of less than 10% is considered de minimis and will *786be acceptable without further inquiry. White v. Regester, 412 U.S. 755, 763, 93 S.Ct. 2332, 2338, 37 L.Ed.2d 314 (1973). Deviations somewhat above 10% may be acceptable if justified by compelling and legitimate interests. See, e.g., Abate v. Mundt, 403 U.S. 182, 184-85, 91 S.Ct. 1904, 1905-06, 29 L.Ed.2d 399 (1971). And, the Court has stated quite clearly that deviations above this buffer range will not be acceptable at all, even if justified by the most compelling and legitimate interests. The Court has not precisely identified the upper range for this buffer category, but does not appear to have approved any plans having a maximum deviation over 20%.11
It should be noted that, in discussing the range of possible deviations, the Court in all of these cases was comparing total population figures. Gaffney, however, tells us that deviation in total population figures is permissible, to some extent at least, because raw population is only an approximation of the number of electors. 412 U.S. at 746, 93 S.Ct. at 2328. It may well be that where, as here, the comparison is between the number of electors, the permissible range of deviation is much narrower.
Even if we apply to this case the ranges established by the Court in cases involving raw population figures, it is clear that the district court's plan falls far outside the permissible range. As far as I am aware, no plan has ever been approved with a maximum deviation of as much as 40%. If I read the Court’s cases correctly, a deviation that large could not be justified even by the most compelling reasons. Nor, do I believe, has the district court even advanced reasons that would permit it to go beyond the 10% de minimis range. Such reasons may exist, but they are not articulated in the record. Four out of the five districts therefore fall outside the acceptable range for purposes of one person one vote.
E. Having concluded that the district court’s plan runs afoul of the one person one vote principle, we arrive at the single most difficult issue in this case: To what extent, if any, this principle may have to give way when it collides with a remedial plan designed to cure the effects of discrimination.
There is, as far as I am aware, little or no guidance on this issue. All prior cases alleging violations of one person one vote involved a conflict between that constitutional principle and various interests advanced under state law. See, e.g., pp. 785-786 supra. Under such circumstances, even if the state is found to have a rational and compelling interest in deviating from substantial district equality, this interest may not justify more than a small range of deviations; beyond that, the state’s interest gives way to the constitutional imperative.
The balance may well be different where, as here, the competing interest is itself grounded in the fourteenth amendment or its derivative, the Voting Rights Act. What seems absolutely clear to me, however, is that the district court cannot simply ignore one person one vote in seeking to create a remedy. The Supreme Court has cautioned that district courts have “considerably narrower” discretion than state legislatures to depart from the ideal of one person one vote, and that “the burden of articulating special reasons for following [policies that would result in a departure are] correspondingly higher.” Connor v. Finch, 431 U.S. 407, 419-20, 97 S.Ct. 1828, 1836, 52 L.Ed.2d 465 (1977). Moreover, “it is the reapportioning court’s responsibility to articulate precisely why a *787plan of single-member districts with minimal population variance cannot be adopted.” Chapman v. Meier, 420 U.S. 1, 27, 95 S.Ct. 751, 766, 42 L.Ed.2d 766 (1975).
At the very least, it seems to me, the district court must make a determined effort to eliminate or minimize the electoral disparities within the districts, consistent with achieving the remedial purposes of the plan. In so doing, I should think the district court would have latitude of up to 20% maximum deviation from the ideal district, providing, of course, that it supplies an adequate explanation of why its purposes cannot be achieved within a narrower range.
What if the district court determines that it cannot construct an adequate remedial plan without going beyond the 20% maximum deviation range? Under one view of the matter, one person one vote would not provide an absolute constraint on the court’s remedial powers, as the competing interest here is not state law — which necessarily takes a back seat to a constitutional imperative — but an interest of equivalent dignity, itself growing out of the same constitutional roots as one person one vote. Under this paradigm, the district court would be allowed, under certain circumstances, to go beyond the 20% buffer allowed by the earlier cases. The district court would have to make very specific findings on how it has sought to achieve substantial equality among the districts and why it has been unable to do so without sacrificing the remedial purpose of the plan. If supported by the record (i.e., if no one comes forward with a plan that can do what the district court says can’t be done), I should think that a much greater deviation from the ideal plan would be permissible, quite possibly as much as the 40% maximum deviation here.
There is, however, another paradigm: A plausible case could be made that the district court gets no greater latitude when it acts pursuant to the Voting Rights Act because its remedial powers are absolutely constrained by the principle of one person one vote. The argument in support of this position grows not out of some hierarchy of values, but out of the nature of the remedial process. A reapportionment plan designed to remedy unlawful discrimination can have one purpose and one purpose only: To put the victims of discrimination in the position they would have enjoyed had there been no discrimination. Here, for example, the object would be to create the type of district that would have existed had the supervisors not continually split the Hispanic core.
Even if we make the most favorable assumptions about what might have been, we cannot conclude that the supervisors would have come up with a district that violated the constitutional constraint of one person one vote. Since we know that, in the normal course of events and in the absence of discrimination, no such district could have been created, no legitimate remedial purpose would be served by creating such a district now. Under this view of the matter, there would be no tension between the court’s remedial power and the principle of one person one vote, and therefore no justification for going beyond the 20% buffer. Even departures beyond the 10% de minim-is buffer could, under this paradigm, be justified only upon a showing that compelling circumstances in the county would, in the absence of discrimination, have resulted in districts of greater than de minimis disparity.12
It is unnecessary to explore this conundrum, however, as it seems absolutely clear that we must remand to the district court on this issue. To begin with, the district court constructed the remedial plan under the mistaken impression that it was constrained by the state law requirement that supervisorial districts be equal in population. It is clear, however, that where state law runs up against a constitutional constraint such as one person one vote, state law must yield. It is most emphatically *788not the case, as the majority suggests, that a district court, in drafting a remedial plan, is constrained by state apportionment law where that law would violate the Constitution.
Remand is also appropriate because the district court was apparently not aware that it was, required to try — if at all possible — to construct a remedial plan that avoided the conflict between the two interests. Since the district court did not try, we do not know whether it is possible to reconcile both interests. A remand is necessary in order to find out. Only if it turns out that an effective remedial plan that also satisfies one person one vote cannot be constructed would I venture an opinion on the difficult question whether, to what extent and under what circumstances the principle of one person one vote must yield when the district court exercises its equitable powers to remedy the effects of past discrimination.
III. Expedited Issuance of the Mandate
Reluctantly, I must also part company with my colleagues in their decision to issue the mandate forthwith. As it is clear from this action that this panel will not grant a stay, we place an unnecessary burden upon the parties, the district judge, our own colleagues and the Justices above us.
I well understand the reason for haste; delaying an election any longer than absolutely necessary should not be done lightly. Consistent with that imperative, we have issued a significant opinion in an important and difficult case about three weeks after submission. No one can justly accuse us of sitting on our thumbs. Were the opinion unanimous, or were I convinced that our differences are relatively trivial, I would go along with expediting the mandate.
But we do not all agree. Moreover, our disagreement goes to the heart of the district court’s remedial plan. Should there be further review, any steps taken by the district court and the parties in implementing the majority opinion would be wasted. The more prudent course, is seems to me, would be to let the parties consider their options in a sober, unhurried fashion, as contemplated by the Federal Rules of Appellate Procedure.
My able colleagues have advanced very compelling arguments as to why the one person one vote rule should be construed as embodying the principle of equal representation. I have suggested that much of the Court’s language and rationale supports the opposite view, that it is the principle of electoral equality that lies at the heart of one person one vote. We are not in a position to resolve this issue, which grows out of a lack of meaningful guidance in a long series of Supreme Court opinions. Yet this issue will have immediate and growing significance as large populations of aliens are taking up residence in several of our largest states.13 The Supreme Court may deem it prudent to take up the issue before large-scale redistricting gets underway in 1991.
Given these considerations, I would preserve the opportunity to have the matter considered in a deliberative fashion, unhurried by the pendency of an election. For better or worse, the election was stayed, which allowed us to consider the case without the sword of Damocles hanging over our heads. I would offer the same opportunity for unhurried deliberation to our colleagues and to any of the Justices who might wish to consider the matter.
IV. Conclusion
This is a fascinating case. It poses many new questions which required the district court to sail into uncharted waters. For the most part, the district court — and the majority — got it right. But close is not close enough when important constitutional rights are at stake. I would order a limited remand for the district court to apply the teachings of Reynolds v. Sims and its progeny.
. The lay reader might wonder if there can be intentional discrimination without an invidious motive. Indeed there can. A simple example may help illustrate the point. Assume you are an anglo homeowner who lives in an all-white neighborhood. Suppose, also, that you harbor no ill feelings toward minorities. Suppose further, however, that some of your neighbors persuade you that having an integrated neighborhood would lower property values and that you stand to lose a lot of money on your home. On the basis of that belief, you join a pact not to sell your house to minorities. Have you engaged in intentional racial and ethnic discrimination? Of course you have. Your personal feelings toward minorities don’t matter; what matters is that you intentionally took actions calculated to keep them out of your neighborhood.
. The district court’s remedy finding No. 5 sets forth the relevant figures for the districting plan it adopted:
District Total White Black Hispanic 9
707,651 25.4 3.5 59.4 - O'
922,180 23.8 50.8 17.1 W
1,098,663 77.0 4.3 13.9 ^1
*780District Total White Black Hispanic Other
4 1,081,089 67.5 4.4 19.7 8.4
5 1,088,388 69.8 6.2 18.1 5.9
TOTAL 4,897,971 55.8 13.4 23.3 7.5
Findings and Order Regarding Remedial Redistricting Plan and Election Schedule 4 (filed Aug. 6, 1990).
. This language is also quoted in Gaffney v. Cummings, 412 U.S. 735, 744, 93 S.Ct. 2321, 2327, 37 L.Ed.2d 298 (1973).
. This language is also quoted in Moore v. Ogil-vie, 394 U.S. 814, 817, 89 S.Ct. 1493, 1494, 23 L.Ed.2d 1 (1969), and Reynolds, 377 U.S. at 557-58, 84 S.Ct. at 1379.
. It is established, of course, that an elected official represents all persons residing within his district, whether or not they are eligible to vote and whether or not they voted for the official in the preceding election. See Davis v. Bandemer, 478 U.S. 109, 132, 106 S.Ct. 2797, 2810, 92 L.Ed.2d 85 (1986) (plurality).
. This language is also quoted in Whitcomb v. Chavis, 403 U.S. 124, 141, 91 S.Ct. 1858, 1868, 29 L.Ed.2d 363 (1971).
. The Court has continued to justify the requirement of equality of populations as a means of assuring that "each citizen’s portion [is] equal.” Morris, 109 S.Ct. at 1438; see also Lockport v. Citizens for Community Action, 430 U.S. 259, 264, 97 S.Ct. 1047, 1051, 51 L.Ed.2d 313 (1977) (“[I]t has been established that the Equal Protection Clause cannot tolerate the disparity in individual voting strength that results when elected officials represent districts of unequal population. ...”).
. In Burns, the ninth and tenth districts contained 28% of Oahu’s total population, yet were entitled to only 6 representatives. The fifteenth and sixteenth districts, on the other hand, contained only 21% of the population, but were entitled to 10 representatives. Burns, 384 U.S. at 90-91 & n. 18, 86 S.Ct. at 1295 & n. 18. Thus, in districts 9 and 10, there was one representative for every 4.67% of Oahu’s total population, whereas in districts 15 and 16, there was one representative for every 2.1% of the population. This deviation of well over 100% — 122%, in fact — far exceeds the population deviations held permissible by the Supreme Court in the line of cases discussed below. See pp. 785-786 infra.
. One's resolve in this regard is put to the test by Brown v. Thomson, 462 U.S. 835, 103 S.Ct. 2690, 77 L.Ed.2d 214 (1983). See id. at 850, 103 S.Ct. at 2700 (Brennan, J., dissenting); note 11 infra.
.My colleagues also rely on the fact that apportionment for the House of Representatives is based on whole population figures. But for reasons explained by the Supreme Court in Reynolds, 377 U.S. at 571-77, 84 S.Ct. at 1386-89, arguments based on the "federal analogy” are "inapposite and irrelevant to state legislative redistricting schemes,” id. at 573, 84 S.Ct. at 1387, and therefore are not particularly persuasive in the context of state and local apportionment cases. Congressional apportionments are governed by section 2 of the fourteenth amendment, which makes total population the apportionment base; it says nothing about state *785apportionments. If this provision were meant to govern state legislative apportionments, the principle of one person one vote, based on a separate part of the fourteenth amendment, would be superfluous.
. The only contrary authority seems to be Brown v. Thomson, as to which it is not clear at all what the relevant deviation was. The only deviation mentioned by the majority and concurring opinions is 89%, which was the degree of deviation of one particularly small county. But the majority and concurrence go to great lengths to assure us that that is not the relevant figure; the two concurring Justices expressed "the gravest doubts that a statewide legislative plan with an 89% maximum deviation could survive constitutional scrutiny_" 462 U.S. at 850, 103 S.Ct. at 2700 (O’Connor, J., joined by Stevens, J., concurring). Because of the peculiar procedural posture of the case, it is hard to tell just what the court viewed as the relevant deviation; it might have been 23%, see id. at 860 n. 6, 103 S.Ct. at 2705 n. 6 (Brennan, J., dissenting), although, for the reasons explained by the dissent, this figure, like the theory of the majority, seems to make little sense.
. The Court has been somewhat vague as to what interests justify departure beyond the 10% de minimis buffer, but the only one clearly identified has been a long-standing and genuine desire to maintain the integrity of political subdivisions. Reynolds, 377 U.S. at 578-81, 84 S.Ct. at 1390-91; Abate, 403 U.S. at 183, 187, 91 S.Ct. at 1905, 1908.
. See, e.g., Suro, Behind the Census Numbers, N.Y. Times, Sept. 16, 1990, § 4, at 4 col. 1.