The plaintiffs launched this challenge to the 1997 districts on April 9, 1997--the same day the ordinance was adopted--claiming, among other things, that the districts constituted an impermissible racial gerrymander and violated the principle of one-person, one-vote.1 On August 15, 1997, the plaintiffs moved for a preliminary injunction. This motion was denied on October 3, 1997. On March 2, 1998, the City filed five separate motions for summary judgment. The district court granted all five motions and issued final judgment in favor of the City on May 7, 1998. Chen v. City of Houston, 9 F. Supp. 2d 745 (S.D. Tex. 1998). This appeal, limited to the one-person, one-vote and racial gerrymander issues, followed.

The plaintiffs maintain that the district court erred in granting summary judgment for the City. They argue that they produced enough circumstantial and direct evidence of the predominance of race in the City's districting decision to allow a reasonable jury to find in their favor. They also claim that the district court erred as a matter of law when it measured the City's compliance with the one-person, one-vote requirement using total population rather than figures that accurately reflected the distribution of potentially eligible voters in the City. Though the issue is extremely close and difficult, after careful review, we have concluded that the plaintiffs failed to meet their evidentiary burden on the Shaw claim and summary judgment was appropriate. We also hold that the use of total population to track the size of the districts does not, under these circumstances, violate the Equal Protection Clause.

We review a motion for summary judgment by applying the same standard as that appropriate for the court below, and in doing so we interpret the evidence in the light most favorable to the nonmovant. Summary judgment is appropriate when the nonmovant fails to demonstrate that there is sufficient summary judgment evidence to allow a reasonable fact finder to find in its favor on all essential issues as to which it would bear the burden of proof at trial. Dutcher v. Ingalls Shipbuilding, 53 F.3d 723, 725 (5th Cir. 1995). However, this review must be understood in the context of the courts' traditional reluctance to interfere with the delicate and politically charged area of legislative redistricting. See Reynolds v. Sims, 84 S. Ct. 1362, 1394 (1964) ("legislative reapportionment is primarily a matter for legislative consideration and determination"); Hunt v. Cromartie, 119 S. Ct. 1545, 1552 (1999) (legislature has the benefit of presumption of good faith when it conducts districting). The Court has clearly indicated that this presumption may impact the assessment of the propriety of summary judgment in a suit challenging districts as racial gerrymanders. See Miller v. Johnson, 115 S. Ct. 2475, 2488 (1995) (sensitive nature of districting should be considered "when assessing under the Federal Rules of Civil Procedure the adequacy of a plaintiff's showing at the various stages of litigation and determining whether to permit discovery or trial to proceed").

In this Shaw challenge, the plaintiffs bear the burden of proving that race was the predominant factor in the City's districting decisions. See Bush v. Vera, 517 U.S.952, 116 S. Ct. 1941, 1951 135L.E.2d 248 (1996). If the plaintiffs can demonstrate that race predominated so that "race for its own sake, and not other districting principles, was the legislature's dominant and controlling rationale in drawing its district lines," the districts will be subject to strict scrutiny. See Miller, 115 S. Ct. at 2486. For a majority of the Supreme Court, the mere fact that race was given some consideration in the districting process, and even the fact that minority-majority districts were intentionally created, does not alone suffice in all circumstances to trigger strict scrutiny. See Shaw v. Hunt, 116 S. Ct. 1894, 1900 (1996) ("Shaw II"); Clark v. Calhoun County, 88 F.3d 1393, 1404 n.2 (5th Cir. 1996) (identifying O'Connor, Rehnquist and Bush dissenters as supporting, with Kennedy reserving the question); Theriot v. Parish of Jefferson, 185 F.3d 477, 488 (5th Cir. 1999). And in the eyes of the author of Shaw, the plaintiffs' burden in establishing racial predominance is a heavy one:

"To invoke strict scrutiny, a plaintiff must show that the State has relied on race in substantial disregard of customary and traditional districting practices.

. . . .

[A]pplication of the Court's standard helps achieve Shaw's basic objective of making extreme instances of gerrymandering subject to meaningful judicial review." Miller, 115 S. Ct. at 2497 (O'Connor, concurring) (emphasis added).

Plaintiffs nevertheless argue that summary judgment is disfavored in Shaw cases. They first argue that the question of the City's intent is inherently a fact-intensive inquiry that is unsuitable for summary judgment, relying on a recent Supreme Court reversal of a grant of summary judgment in a districting case. See Cromartie. This ignores, however, the fact that in Cromartie the district court had granted summary judgment for the plaintiff. The Court took care to stress that summary judgment was inappropriate not only because of the factual nature of the intent requirement, but also because the nonmovant did not bear the burden of proof on the issue. See id. at 1552 ("Summary judgment in favor of the party with the burden of persuasion, however, is inappropriate when the evidence is susceptible of different interpretations or inferences by the trier of fact.") (emphasis added). The Court also relied on the traditional presumption that the legislature acted in good faith while districting. Id. Here,neither of these factors are present. The plaintiffs here bear the burden of persuasion, and the presumption of legislative integrity adds to, rather than lessens, their burden facing summary judgment.

The plaintiffs next argue that summary judgment is inappropriate because the case is one of mixed motives--while the City has proffered justifications for its actions grounded in traditional districting principles, it has also conceded that race was a factor in its decisions. The plaintiffs argue that the Court has indicated that a more searching inquiry is needed in such cases, and summary judgment is thus inappropriate. Bush v. Vera, 116 S. Ct. 1941, 1952 (1996) ( in a mixed motive case a "careful review is, therefore, necessary to determine whether these districts are subject to strict scrutiny"). This position misconstrues Bush. The Bush court made this statement to distinguish mixed motives cases in which the districting body presented credible evidence of alternative justifications from those in which no valid traditional districting purpose was given and racial intent was obvious. See id. (distinguishing with cf. citation mixed motive case from Miller); Miller, 115 S. Ct. at 2485 (noting that district court had found evidence of predominance of race was overwhelming "and practically stipulated by the parties involved"). Far from indicating that summary judgment should be disfavored, the discussion the plaintiffs highlight demonstrates the gravity with which the court approaches facially valid defenses proffered by the legislature in thedistricting area. The City of Houston's presentation of valid evidence of nonracial intent, which transforms the case into one of mixed motives, advances rather than hinders its case for summary judgment. We will review the district court's grant of summary judgment using the traditional criteria, and take into account both the presumption in favor of a legislature's good faith and a plaintiff's burden of proof.

II. Circumstantial Evidence of Racial Motivation--The Appearance of the Districts

A plaintiff may demonstrate that race predominated in a districting decision by introducing circumstantial evidence of the district's shape and demographics. See Shaw II, 116 S. Ct. at 1900 (1996). The shape alone of some districts may be so bizarre and irregular that their creation may only be interpreted as "an effort to segregate the races for purposes of voting, without regard for traditional districting principles." Shaw, 113 S. Ct. at 2824; Cromartie, at 1549 n.3 (some highly bizarre districts give rise to inference that race predominated more strongly than the inference that factors such as politics created the distortion). When shape alone is not determinative, plaintiffs may still prove their circumstantial case by demonstrating that the districts are sufficiently bizarre in relation to racial demographics and population densities. See Miller, 115 S. Ct. at 2489. The objective case against a district may also be affected by an analysis of whether the district adheres to traditional districting principles other than compactness, such as political considerations and fidelity to administrative boundaries. See Bush, 116 S. Ct. at 971.

In a very general sense, Houston's nine single member council districts can be described as being arranged like pieces of a pie, each district radiating out from the center to the edge of town. The two exceptions are District H, which does not touch the city's external borders due to a thin bridge between two halves of District B, and District E, which does not approach the city center. Were the City a congressional district carved out of the surrounding suburbs, its own shape would likely provide ample fodder for a Shaw claim. The City engulfs independent incorporated areas, leaving two holes in its west half. It also includes three detached "islands" to the north of the City proper--two of them quite large--that are only tenuously connected to the "mainland" by narrow and unpopulated corridors. In addition, it includes several peninsulas jutting out into the countryside, the most outlandish of which extends in a snakelike fashion eastward out of District E between the IH-10 Freeway and the La PorteHighway. In short, as a glance at the map will readily confirm, the City has vastly irregular borders and does not even remotely approach a mathematical ideal of compactness.

But plaintiffs do not challenge the shape of the City of Houston, and the City is not in fact a congressional district. The City's highly irregular borders are simply a given that the Council was forced to deal with as best it could in crafting the districts challenged here. Because of the constraint imposed by the City's external borders, Houston's city council districts will never enjoy the pleasing symmetry of, say, Nebraska or Kansas (to say nothing of Colorado or Wyoming). The City's various peninsulas and protuberances, and in particular its archipelago of detached islands, have to be placed somewhere. While this might at first glance leave particular districts open to charges of being noncompact, these breaches of that ideal are unavoidable under the circumstances. Recognizing this, when drafting its guidelines for the redistricting, the city's geographic goal embraced functional as well as purely spacial compactness considerations.2

This rather obvious point seems to have escaped both the plaintiffs and their expert witnesses. The plaintiffs take care to lift segments of the City's districts and compare them to elements of Shaw, Reno, and Miller, but fail to point out this context. Thus, for example, the plaintiffs argue that the City's districting plan used narrow, unpopulated connectors to link District B with the Houston Airport area, and that District E has a snaking protuberance extending far to the east under the challenged plan. While this is surely correct, and leads to an immediate analogy to the Shaw district, it is absolutely irrelevant. Were Houston to purge race utterly from its districting process by handing over the task to a computer programmed to maximize purely spacial compactness, these deviations would still be present. The same error infected the expert opinion relied on by plaintiffs to demonstrate lack of compactness and bizarre shape. Plaintiffs' expert, Dr. Maggiotto, performed the type of mathematical analysis cited with approval in several cases and concluded that the City's districts were not compact.3 However, Dr. Maggiotto failed to take into account that substantial deviation from absolute mathematical compactness was inherent in the City's current form. Accordingly, we reject his conclusion. Even on a motion for summary judgment, we are not required to take heed of an ill-reasoned expert opinion. See Matsushita Electric Industrial Co. V. Zenith Radio Corp., 106 S. Ct. 1348, 1360 n.19 (1986).

This general failure of the plaintiffs does not, however, foreclose their case. They point to several districts which have borders that deviate from the Euclidean ideal in a manner not entirely traceable to the City's external borders. In particular, they highlight districts B, H, I, and E.4 District B, which has an African-American majority, consists of one "island" around the Houston airport, and two distinct areas on the "mainland" that are divided by District H and are connected only by a thin strip of land along H's northern border. The eastern half of the District B plungesat one point into District I, and also has its pattern broken by the intrusion of an arm of District H. District H, while possessing a highly compact substantially rectangular core, has two "arms" extending to the west and southeast. The southeast arm curls around the borders of districts B and I in an ungainly fashion and terminates in a small, uneven projection extending a short distance back to the west, into District B. The far more regular western arm of District H extends for a fair distance into District A, and at one point tapers sharply to merely point continuity before expanding again. District I by comparison is far more compact, although its eastern and western sections are separated to a certain extent by District B. District E consists of a southeastern portion that would be fairly compact but for the City's borders. However, it also includes the large island in the Lake Houston area to the extreme northeast of the city, with which it shares a connection only by a narrow eight-mile corridor. The nature of this connection cannot be helped, as it is a product of the City borders. As the plaintiffs stress, however, the Lake Houston island is closer as the crow flies to Districts B and I than to the balance of District E. In previous plans, the portion of this island then within the City was in fact placed in District B.

Certainly, these four districts--B, H, I and E--are not perfectly compact. And since District B is a predominantly African-American district, Districts H and I contain Hispanic majorities, and District E is mainly white or "Anglo," the shape of these districts is not so regular as to preclude the plaintiffs from showing that race predominated in the drawing of a portion of some of the extremities of these districts. But these districts are not the type of sprawling monstrosities that have previously been found to violate Shaw by the Supreme Court. Indeed, Bush v. Vera discussed districts that substantially overlapped with the contested districts H, I and B. A quick review of the map in Bush will demonstrate that the City was far more faithful to the ideal of compactness than the Texas Legislature was in drafting the Bush plan. In Bush, legislators attempting to create Houston area seats for both Blacks and Hispanics effected this goal by selecting block-sized packets of Hispanic or African-American voters and linking them each into their respective district. The end result created two of the three least regular districts in the country, comparable to a jigsaw puzzle in which the pieces could not be removed. See Bush, 116 S. Ct. at 1958. In Bush, in the area at issue here, an already somewhat ungainly tentacle of the African-American district contained uneven veins of loosely interconnected Hispanic blocks that were placed in, and attached tenuously to, the Hispanic district. Not only does the City's districting of the same area fail to even approach that level of non-compactness, the record indicates that the Council specifically rejected a proposal by one of its members to redraw its districts to mimic the plan rejected in Bush on the grounds of noncompactness. In addition, we note that the City successfully resisted pressure from private litigants to draft a plan containing four majority-minority districts--something that almost certainly would have created a more bizarre plan. See Campos, 113 F.3d 544. Race was clearly not the only factor driving the City's districting, because it avoided these extremes.

While we certainly do not mean to suggest that the extremes of Bush or Shaw are the minimum threshold for a successful suit, we simply cannot find that the challenged shape of the districts here offers adequate evidence that race predominated in the districting. This is especially true since some of the specific departures from compactness that the plaintiffs highlight are not as dramatic as they might initially appear. First, the plaintiffs attempt to paint the odd southeastern arm of District H as an attempt to maximize Hispanic numbers in H while avoiding the inclusion of African-Americans. But this argument ignores the fact that District H's arm borders not only "Black" District B, but alsoDistrict I--the other majority Hispanic district in Houston. Had the City's sole real concern been maintaining the racial purity of these districts, that goal could have facially just as easily been accomplished by placing H's southeastern arm in District I, in exchange for the placement of District I's northwest ern corner in District H. This configuration would have appeared far more compact and regular, since the arm of H blends naturally into the north of I. And since the relevant borders of the arm itself would not have altered, the African-American population of B would have been unaltered and the percentage of Hispanics in both H and I would have remained roughly equivalent. Careful demographic analysis of the populations in these disputed areas might disprove this hypothesis, but the plaintiffs failed to even attempt to provide such data. The evidence does not tie the particularly unusual feature of H's borders to race to the exclusion of any other real concerns.

Second, the usage of a narrow connector to link the two sections of District B separated by District H is also not as facially objectionable as it first appears. The plaintiffs argue that District B's odd shape is explained by the City's desire to carve out a Hispanic majority for District H, and continuity in B was accordingly sacrificed, with the narrow connector a mere token concession to the principle. However, when the area of District B as a whole, including the dead space not included in the City, is taken into account, this picture is far from clear. Because a large section of District B is a detached island, B is in some respects missing its center. The geographical heart of the district is located outside the City limits. The northern island of B and its two mainland halves would constitute a facially unobjectionable compact body if the intervening unincorporated territory were included. If one takes this into account, and views District B as an attempt to reflect a coherent geographical neighborhood in the north of the City that happens to be broken up by unincorporated dead space, the outline of the district becomes much less offensive. Rather than being a racial subterfuge, the strip of territory connecting the incorporated halves of B can be properly seen as marking the central border of a coherent and relatively compact district whose outline is only clouded by the City's irregular borders.

Lastly, the presence of the detached Lake Houston area in District E offers less support for the plaintiff than it otherwise might. Concededly, the placement of this area leads to the creation of an ungainly district that sprawls from the northern to the southern extremity of the City. And this northeastern island is closer to District B, so it would seem to do violence to the compactness concept to put it anywhere else. However, this must be placed in the context of the City's most unusual outline. When a districting body has the opportunity to create truly compact districts and chooses not to, the harm is obvious. But when, as here, spacial compactness is simply impossible, further departure from the principle seems less jarring. Wherever the island was placed, its representative would have to travel several miles outside of the City proper, and voters in that district would have had the odd sensation of being thrown together with an area with which they shared only a technical connection. Given that the City's design had already decisively severed the island from its neighbor, its placement in a district that was a few miles farther away seems less offensive than it would otherwise. A judicial analogy may serve to clarify this point. We would find it odd indeed if Congress decided to place Washington State within the realm of this Circuit when the opportunity to keep it within the Ninth Circuit was present. But it is far less troublesome that Puerto Rico has been assigned to the First Circuit rather than the Eleventh (or the Fourth). The placement of Kingwood and its environs is discussed in more detail below, but facially we find that its separation mitigates to alarge extent the effect of its placement in District E.

In short, when viewed properly, the geographic boundaries here do not in and of themselves make out a strong circumstantial case under Shaw. That does not end our inquiry, however. As Miller indicated, a district that is not facially outrageous may be objectively revealed as a gerrymander by linking its shape to demographics and other factors. In Miller, the Court conceded that the district's shape, while hardly perfect, was not sufficient standing alone to lead to the conclusion that it was bizarre in a Shaw sense. It then turned to a closer analysis of the geography in light of the district's demographics. It noted that the district's shape became far more troublesome once it was recognized that the core of the district was relatively unpopulated, while the carefully drawn exterior borders reached out to engulf urban African-American population centers. Viewed in light of this evidence, a shape that initially merely raised eyebrows was revealed to present overwhelming circumstantial evidence of a constitutional violation. See Miller, 115 S. Ct. at 2489 ("Although by comparison with other districts the geometric shape of the Eleventh District may not seem bizarre on its face, when its shape is considered in conjunction with its racial and population densities, the story of racial gerrymandering seen by the District Court becomes much clearer.").

Here, as in Miller, we are presented with districts in which deviations from the compactness ideal are present but are not overwhelming. But unlike in Miller, the plaintiffs have failed to present us with specific demographic data that would clarify the racial nature of these distortions. The plaintiff's expert, Dr. Weber, merely produced demographic data covering the districts as a whole. He did not give us a demographic analysis of the specific features--the arms of H, the connector section in B, and the northeast island in E--that are at issue in this case. His report did not show that, for example, the arms of District H contain most of the district's population, and H's compact core is relatively unpopulated. Nor was there even data demonstrating that the western arm of H is designed to capture a specific pocket of Hispanics in order to maintain Hispanic supremacy in the district. Based on a map of Hispanic population dispersion in the record, it appears this arm includes several areas without, and excludes some areas with, high ratios of Hispanics. The general, over-broad data Dr. Weber used supports only his conclusion that these districts could be made more heterogenous. And this is hardly relevant--of course if one solely focused on increasing racial heterogeneity one could improve this factor, just as one could increase homogeneity by adopting the pattern rejected in Bush. But districters are not bound--or allowed--to sacrifice traditional districting concerns to meeting quotas of diversity, just as they are not allowed to do so in order to meet quotas of racial concentration.

Plaintiffs also argue that the fact that the number of racially homogenous precincts increased in the 1997 plan demonstrates that the City's districting was predominantly linked to race. Standing alone, however, this fact does nothing to support the plaintiffs' case. For reasons of their own, populations occasionally shift within urban areas. If members of one racial group chose to move out of certain areas, the numbers of racially homogenous precincts would increase. However, this phenomenon would not be tied to the actions of the districting parties-they could keep the borders of the districts static and the same effect would be seen. Indeed, according to the plaintiffs' version of events this is precisely what occurred--they criticize the 1997 plan for substantially maintaining prior boundaries. Without tying this evidence to specific choices by the City, and in particular to the discrete noncompact elements of the plan that they challenge, this evidence fails to aid the plaintiffs' case. Ultimately, the plaintiffsfailed to demonstrate by use of demographics, as was done in Miller, that the specific departures from compactness that it challenges are the product of race rather than other factors.

However, compactness is not the only traditional districting principle. Other factors, such as the protection of political incumbents, the maintenance of administrative units, and the reflection of communities of interest, play a part in the districting process. By showing that the design of a district violates these other traditional districting principles, the plaintiffs support their circumstantial case.5 The districts clearly adhered to at least one traditional districting principle by maintaining the integrity of the relevant administrative unit, voting precincts (although as discussed below plaintiffs argue that these units are themselves tainted). However, the plaintiffs argue that the districts do not serve the traditional districting principle of reflecting communities of interest. In the more egregious examples of gerrymandering, the violation of this traditional principle will be facially obvious. Thus the district challenged in Shaw could be safely assumed not to have been driven by communities of interest, since it "winds in snakelike fashion through tobacco country, financial centers, and manufacturing areas" in an attempt to gather up "enclaves of black neighborhoods"and "even towns are divided." Shaw, 113 S. Ct. at 2821.

But when presented with closer questions, the Court has focused on the specific features of a district that cause it to depart from compactness in assessing whether they are unjustifiable on traditional grounds. Thus in Bush, where the State attempted to justify its decision-making on the analogous traditional ground of partisan politics--and there was indeed a correlation with politics when the districts were looked at as a whole--the Court found it significant that "the maps reveal that political considerations were subordinated to racial classification in the drawing of many of the most extreme and bizarre district lines." Bush, 116 S. Ct. at 1957. The fact that the State was willing, in the course of extending a noncompact tentacle engulfing a pocket of African-American voters, to include a large number of Republican voter tabulation districts in a congressional district ostensibly designed to maximize Democratic power was clear evidence of the subordination of politics to race. Id. And in Miller, the demographic data was heavily reinforced by an analysis of how specific choices violated communities of interest. Because the challenged district reached out to include African-American pockets in several entirely separate urban communities, and linked them together with a sparsely populated and wholly rural core, the "social, political and economic makeup of the Eleventh District tells a tale of disparity, not community." Miller, 115 S. Ct. at 2484.

Here, the plaintiffs have failed to link the specific departures from compactness that they highlight to the violation of traditional districting principles. They first claim that several neighborhoods were split in the 1997 plan. But the City introduced evidence that more neighborhoods were reunited than were split by the plan, and the plaintiffs have not challenged this assertion. Moreover, the plaintiffs failed to show where the allegedly split neighborhoods were. While it may be unfortunate that a predominantly white neighborhood was split between two white districts, the existence of such a districting faux pas may well have nothing to do with the racial issues that concern us here. The plaintiffs' evidence of neighborhood splitting is not helpful to their case.

Plaintiffs also tendered the opinion of Dr. Maggiotto, who analyzed various purportedly relevant statistical indicators ofcommunities of interest.6 His report claims that, in the City as a whole, race varies less within the districts than do his chosen indices of community. This evidence, it is argued, demonstrates that race predominated in the City's districting. However, the plaintiffs again use too blunt a statistical tool. Socio-economic variation within districts is almost unavoidable, particularly when, as here, the constituent districting units are fairly large--approaching 200,000 persons. Not many, if any, rectangles of this size randomly deposited on a map of Houston could be expected to capture an essentially homogenous socio-economic group. At least when relatively large districts are involved, the mere existence of socio-economic variations within the district is not probative. To provide valid circumstantial evidence of the predominance of race in the City's districting, then, Dr. Maggiotto had to tie his data to specific districting decisions, or the plaintiffs had to place his data in context by showing that the variations from available normal compactness were likely caused by racial factors. This they failed to do. They did not show--or even claim--that, for example, the southeastern arm of District H extends outward to reach a pocket of wealthy Hispanics who share nothing other than race with the bulk of the district, the way that the Atlanta extremity of the Miller district shared no commonality with the district's rural core other than race. Nor did they show that the western arm of H is designed to reach a pocket of Hispanics at the cost of including on the way neighborhoods that share nothing with the rest of that district, the way that Texas reached out to a pocket of African-Americans at the cost of including Republican districts in Bush. Indeed, the plaintiffs never analyzed or even referred to the socio-economic composition of the specific areas created by the City's departure from geographical compactness, relying instead on broad statistics covering the entire area. See Bush, 116 S. Ct. at 1957 n.* (criticizing dissent for relying on statistics showing political orientation for district as a whole, rather than concentrating on border segments). Had they done so, perhaps our decision today might have been different. But bearing in mind that it was the plaintiffs' burden to produce evidence that race predominated, and also remembering the presumption in favor of legislative integrity, we cannot say that the shape and demographics of the districts here provided adequate circumstantial evidence of the predominance of race to prevent summary judgment.

Plaintiffs also argue that they have sufficient direct evidence that the City's districting was predominantly driven by race. They focus on two major points. First, they argue that the major decision made during the 1997 districting-that the newly annexed area of Kingwood should be placed in District E rather than District B--was predominantly influenced by race. Second, they argue that the 1997 plan substantially maintained the borders of previous plans, and that those borders were set by a process in which race--specifically, the desire to create two black and two Hispanic single member districts--predominated. The City concedes, as it must, that race--in the form of a desire to comply with the Voting Rights Act--was a factor in all the plans. But aswe have stated before, a majority of the Supreme Court does not believe that the mere presence of race in the mix of decision making factors , and even the desire to craft majority-minority districts, does not alone automatically trigger strict scrutiny. See Clark, 88 F.3d at 1404 n.2; Theriot, 185 F.3d at 488.

Before touching on plaintiffs' most heavily pressed intent evidence, we briefly dispense with their argument that racial intent can be inferred from the City's use of precinct lines. Plaintiffs assail the district lines because they hew to precinct lines. Conformity to such administrative boundaries is ordinarily seen as a virtue, not a vice, in the districting process. See Bush, 116 S. Ct. at 1953 (noting maintenance of some county lines showed traditional factors were not entirely ignored). And plaintiffs have not identified--and there does not appear to be--a larger, more race-neutral, administrative unit that could have been practically deployed. Cf. Miller, 115 S. Ct. at 2490 (noting that plan split several counties, and thus violated traditional districting principles-use of precincts could not mitigate this fact in congressional districting). But the plaintiffs stress that the City had access to racial data on the precinct level. They attempt to analogize this situation to Bush, where the State used a computer system that had racial data available on the block level. The fact that the plan ultimately adopted split precincts and used block units instead was found to offer compelling proof that race has predominated in the districting. But the key factor in Bush was not the mere availability of racial data. Rather, it was the abandonment of a traditional districting unit, the precinct. Since this step violated traditional districting principles and caused substantial administrative problems, and since given the fact the only additional precision the computer provided at the block level was racial, the obvious inference was that race drove the decision. See Bush, 116 S. Ct. at 1959. Here, in contrast, the City adhered to the traditional--and apparently the only--administrative boundaries readily available. The plaintiffs complain that this is irrelevant, since the size of the precincts was deliberately shrunk to allow more precise racial gerrymandering. However, the City of Houston does not set precinct size, and the plaintiffs have not tied the changes in precinct sizes to any actions of the City.

The major change made by the City in 1997 was driven by the City's decision to annex Kingwood, an overwhelmingly white (Anglo) planned community with a population of around 40,000 people that is attached to the detached Lake Houston area. The decision to annex Kingwood is itself not before this Court. Rather, the plaintiffs complain that Kingwood--and with it the rest of the northeastern island around Lake Houston-- was placed in District E. Previously, this island had been attached to District B. They claim that they have direct evidence that race predominated in this decision--specifically, that the City's overwhelming motivation in doing so was avoidance of potential section 5 retrogression liability. Because Kingwood has a large population, and because District B was close to the one-person, one-vote population limit, placing Kingwood in B would have forced the City to move areas on the borders of B into the surrounding districts. And because the only district adjacent to B with a white (Anglo) majority--A--was already relatively overpopulated, moving areas from B to A would have in turn created more spillover effects, resulting in the disruptive necessity of redrawing several districts. Placement of spillover from B into Districts H or I would initially seem more attractive, since those districts were underpopulated. However, the addition of African-American voters displaced from B would seemingly have eliminated the Hispanic majorities in either or both of these districts. The City believed that this might have triggered retrogression concerns under section 5 of the Voting RightsAct.7 Placement in District E, by comparison, allowed for the smooth integration of Kingwood. Spillover and one-person, one-vote concerns were minimized, with only relatively minor border changes made necessary.

The evidence clearly indicates that the City's decision to place Kingwood in District E was partially driven by the fear of retrogression. Its section 5 filings and the statements of both the drafter of the plan and the council members involved establishes this, and the City does not contest it. Indeed, it argues that this concern would allow it to survive strict scrutiny review. And to a certain extent--as discussed above, the issue is not as pressing when a detached island is concerned--compactness was sacrificed in the placement of Kingwood. Kingwood is closer to District B than it is to District E. However, a review of the record indicates that the Council had before it at the time it made its decision to place Kingwood in District E two other legitimate grounds for justifying its decision. First, the City wanted to preserve communities of interest. Second, it wanted to minimize disruption in the run-up to the expected major redrawing of districts following the 2000 census. On this record, we cannot say that there has been an adequate showing that race predominated over these concerns.

The most powerful alternative justification for the City's actions is the nature of Kingwood and it greater community of interest overlap with portions of District E. The Court has noted that concern about communities of interest is a valid traditional districting tool that may serve to deflect an inference that race predominated in districting. See Bush, 116 S. Ct. at 1954 (evidence that "district lines were drawn in part on the basis of evidence (other than racial data) of where communities of interest existed might weaken a plaintiff's claim that race predominated in the drawing of district lines."). Here, transcripts of the series of council debates that led to the final vote on the districting plan demonstrate that members of the Council discussed and evaluated data that indicated that Kingwood's communities of interest overlapped with District E, but were wildly at variance with District B. The fact that this material was available to the Council at the time they were deliberating on it serves to distinguish this case from others in which alternative justifications have been raised after the fact and only hypothetically motivated the districting body. See Shaw II, 116 S. Ct. at 1902 n.4 (purported justification must be actual purpose in adopting plan and supported in the evidence); Bush, 116 S. Ct. at 1955 (disregarding community of interest justification because data was not before the legislature in an organized fashion).

In the Council debates, members of the Council referred to both anecdotal and statistical evidence demonstrating that the average income and quality of housing in Kingwood were much higher that those found in District B, while the percentage of persons on public relief and the occurrence of illiteracy was much smaller. Moreover, the members of the Council did not limit themselves to general discussions of socioeconomic indicators. They pointed to specific factors that demonstrated that District B's community concerns varied from those in Kingwood. Because Kingwood is a planned community, it has excellent infrastructure. In comparison, the infrastructure in District B was referred to as among the worst in the City. This wouldraise the concern that the residents of Kingwood's desire and need for municipal improvements would diverge from those of other residents of District B. It was also stated that Kingwood had privatized several municipal services, and paid for services using regressive user fees rather than more progressive property taxation. This difference would obviously lead to a divergence between the two areas' approaches to services and taxation.8 In comparison, District E was said to be far more congruent with Kingwood than District B. Its infrastructure and average income were much closer to Kingwood's. It also contained planned communities that shared Kingwood's special concerns regarding services and taxation. In light of the greater similarity between the two areas, in the eyes of the members voting for the plan it made sense to place Kingwood in District E.

Significantly, opponents of the plan--and two Council members, as well as the plaintiffs' lawyer in this case, spoke out in some depth against it before the vote--never challenged these assumptions. At no point were the Council members' characterization of Kingwood's socio-economic status, infrastructure, and services called into question in any way. The closest an opponent of the plan came to undermining the majority's community of interest assumptions was a discourse by one of the members who voted against the plan. He argued that communities of interest were inherently linked to geography, and thus could not bind Kingwood to District E. In its strong form, this statement would make communities of interest a mere subset of geographic compactness. In its weak form, it is generally correct and a useful observation to guard against abuses of the community of interest rationale. Here, however, this reasoning is inapplicable. Since Kingwood was separated from the main body of both Districts B and E, it was never likely to share a sense of geographic neighborliness with either district. The question was not one of placing Kingwood in either District E or an adjacent community linked in the eyes of the populace with Kingwood. Instead, the issue was which of two geographically isolated areas Kingwood and its surrounds should be shoehorned into--a district in which it had nothing in common or one with which it shared at least some points of community concern.

Nor have the plaintiffs pointed on their appeal to any evidence that undermines the community of interest justification. As noted above, their arguments and evidence regarding communities of interest are fatally flawed by the broad level of generality they adopt. Here, this weakness is especially glaring. Plaintiffs have not presented data about Kingwood at all, let alone introduced evidence that undermines the arguments made by proponents of the plan at the time. Since Dr. Maggiotto did not separate out Kingwood, his data cannot be used to undermine the City's justification here--there is no evidence that factors such as occupations or educational level create a significant point of commonality between Kingwood and District B that mitigates the divide created by raw income and other factors. Indeed, Dr. Maggiotto's evidence largely confirms the City's analysis. It shows that District B was the second poorest district in the City, while District E was the second richest, and contains the lowest variation of income. At most, plaintiffs have shown that all of District E is not exactly congruent with Kingwood. They have not shown, and have never claimed, that Kingwood was more similar in regard to the factors discussed to District B than it was to District E. And they never attempted to introduce, by expert opinion or otherwise, evidence tending to show that this lack of similarity was overcome by the presence of commonalities in nonquantifiable indicia ofcommunities of interest.9 In short, this is not a case where a decision is justified by a "mere recitation of purported communities of interest" unproven by facts in the record. Cf. Miller, 115 S. Ct. at 2490.

Further support for the City's position is provided by the special nature of the decision of where to place Kingwood. Not only was the council presented with the awkward task of placing a disconnected, newly-acquired parcel of land into a settled districting scheme, it also did so against the looming background of the upcoming census. As a result of the 2000 census, the City will in the near future receive significantly better data about its population. A report prepared for the Council by a local university professor, Dr. Alford, indicated that the City would be required to engage in a major overhaul of its districts in light of this data, and that a similar effort in 1997 based on soon-to-be obsolete census figures would be inefficient. The City evidently has followed a practice of confining major changes to the districting that follows the census. See Campos v. City of Houston, 968 F.2d 446, 449 (5th Cir. 1992) (discussing replacement of old plan "used during the 1980s" in wake of 1990 census). Inlight of this, it is hardly surprising that the City would be reluctant to place Kingwood in an area which would require major alterations of the district plan. Several Council members spoke to this effect. By placing Kingwood in District E, the City minimized the spillover effect mandated by one-person, one-vote concerns, and the corresponding mass redrawing of districts that would ensue. Under the special--if not wholly unique--situation created here by a major annexation located so soon before the next census, this would seem a valid concern.10

The plaintiffs attempt to undermine this evidence by arguing, in essence, that the Council's debates were mere smoke and mirrors concocted in anticipation of precisely this lawsuit. Certainly, the debates evidence an intense awareness of the possibility that the districting would be subject to a Shaw suit. The first day of the debate began with legal counsel explaining the nuances of Miller and Bush in great detail to the Council. However, this record does not support a conclusion that the whole debate was essentially scripted. The voluminous testimony exhibits meaningful disputation, and several proponents of the plan used language and took positions that are not consistent at all with the hypothesis that the supporters of the plan were carefully coached. Given the presumption in favor of legislative good faith granted in this area, we cannot say on this record that the mere fact that the Council was aware of the meaningful possibility of this litigation suffices to create a fact issue as to their statements being mere pretexts for concealed racial motivation. We cannot say that a fact issue has been created regarding the predominance of race in the placement of Kingwood. The decision was independently justified by several factors besides race at the time it was adopted, and the force of the traditional concern that was sacrificed--compactness--was diminishedby the context of placement of a detached island.

B. Continuation of a racially motivated districting structure

Plaintiffs also contend that the City violated Shaw by failing to make substantial changes in a districting structure that was originally designed in a manner that race predominated over traditional districting concerns. The sins of past Councils, and the Department of Justice (DOJ), thus are claimed to haunt the current Council's asserted ratification of their work. The district court summarily rejected this argument, reasoning that the constitutionality of the prior plans had never been challenged, and the intent behind their creation was thus not properly before the court. This would appear to have been overly hasty. In the context of Voting Rights Act suits, evidence that impermissible racial intent had tainted the plan upon which the challenged plan was based has been allowed, even when enough time has elapsed for a substantial degree of familiarity and political reliance to emerge. See Garza v. County of Los Angeles, 918 F.2d 763, 768-69 (9th Cir. 1990), cert denied 111 S. Ct. 681 (1991). The mere passage of time cannot extinguish entirely the taint of racial discrimination. See Hunter v. Underwood, 105 S. Ct. 1916 (1985) (discriminatory intent behind 1901 disenfranchisement provision coupled with continued discriminatory impact relied on to invalidate statute). And in a context like the one before us, where the City redistricts relatively frequently, adoption of the City's position would seem problematic because it would give potential challengers an incentive to launch more frequent lawsuits. If the potential plaintiffs did not sue, they might find that the passage of other plans had sanitized the intent embodied in the prior plan.

We first note that while race clearly played an important role in the earlier districting efforts, the quality of evidence introduced by the plaintiffs in regard to the earlier efforts (as well as respecting the 1997 ordinance) falls far short of that introduced in the cases examined by the Supreme Court. As discussed above, the objective, circumstantial evidence of the predominance of race is far weaker than in those cases. As for direct evidence, in the Shaw cases not only was the end product of DOJ intervention a monstrosity that massively violated traditional districting principles, but the State also openly stated in its section 5 filings that race was its dominant concern rather than just a factor. See Shaw II, 116 S. Ct. 1901 (preclearance submission announced creation of minority districts was State's "overriding purpose"). A similar direct admission of intent was present in Bush. See Bush, 116 S. Ct. at 1957 (testimony of state officials in related litigation claimed that race was the primary consideration in districting). And in Miller, where--as here--the district's shape was not itself extremely bizarre, the evidence that race trumped other concerns was conclusive. There, the DOJ rejected proposed plans not once, but twice, and ultimately made it clear to the State that it would only accept a result that mirrored a plan proposed by a third party, the ACLU. Ultimately a plan designed around this framework was adopted, as legislators who voted in its favor testified,11 simply because the State felt it had no other choice. Obviously, the ACLU lawyer who drew up the outlines of the plan challenged in Miller could have no firm understanding of, or respect for, the subtleties of traditional districting. See Abrams, 117 S.Ct. at 1933(State "adopted the Justice Department's entirely race-focused approach to redistricting"); Johnson v. Miller, 864 F. Supp. 1354, 1368 (S.D. Ga. 1994) (lower court discusses role of ACLU drafter). The inference was thus automatic that race had not only predominated over, but had indeed almost entirely extinguished, traditional districting concerns.

Here, in contrast, the only evidence that the plaintiffs produced is the raw history of DOJ intervention, which contains no brazen admission that race predominated and no details regarding a pattern of DOJ pressure.12 Plaintiffs relied on the fact that in 1991 the City's original plan was denied preclearance, and that while the plan the City drafted in response creating additional concentrations of minorities was never implemented, it formed the template for the 1993, 1995, and 1997 districtings. Based solely on these facts and the section 5 filings that accompanied the plans (and demographic evidence that, as we discussed above, is insufficient to sustain plaintiffs' case), the plaintiffs' expert Dr. Weber opined that race had predominated in all of the City's districting since 1991. Certainly, this material establishes that race was a factor in prior districtings, and that the form of those districts was largely reenacted in 1997. It also establishes an intention on the part of the City to establish minority-majority districts. And, a glance at the district maps verifies that the City's original plan was more compact than the ones subsequently adopted.

But the City has always conceded these points, and as we discussed above the sacrifice in compactness was not extreme. And this Court has interpreted the Supreme Court's current position to include a majority in favor of Justice O'Connor's statement in Bush that the intentional creation of minority-majority districts will not in and of itself trigger strict scrutiny. See Bush, 116 S. Ct. at 1951; Clark, 88 F.3d at 1404 n.2; Theriot, 185 F.3d at 488. The plaintiffs' burden was thus to show that race predominated over traditional districting concerns, not merely that the City attempted to craft minority-majority districts and that there was some but not massive reduction of one such factor--compactness. Here, all that has been demonstrated is that the DOJ aggressively forced a consciousness of race and the Voting Rights Act upon the City. The City then abandoned its initial plan--one presumably grounded firmly in traditional districting principles-and ultimately adopted a system that met the DOJ's requirements. Were districting an exact science, in which the application of traditional districting principles always produced one objectively optimal solution, this would perhaps settle the question--abandonment of the optimal traditional plan would demonstrate subordination. But districting is hardly a science. It would seem obvious to us that there is more than one way to draw a district so that it can reasonably be described as meaningfully adhering to traditional principles, even if not to the same extent or degree as some other hypothetical district.

If, in the wake of the DOJ's intervention, the City had returned to the problem and properly considered race as a factor alongside traditional districting principles, and the result they reached had satisfied the DOJ without substantially abandoning those principles, race would not automatically have predominated. For example, the City could have returned to the drawing board and generated a plan that increased the recognition of communities of interest and satisfied the DOJ's racial concerns.13 The fact that compactness wassomewhat lessened in the process would not establish that strict scrutiny was required, because race might not have predominated over communities of interest. Thus abandonment of an original plan by itself does not demonstrate that race predominated--the adopted plan may have also been the product of and substantially consistent with traditional districting principles. The plaintiffs failed to offer any evidence that this did not in fact occur. As discussed above, the plaintiffs failed to show specific areas in which communities of interest were subordinated to race, and they never claimed that the original plan was a more adequate reflection of that traditional districting principle than those that followed. The plaintiffs' expert, Dr. Weber, conceded that he did not take into account the City's desire to reflect communities of interest, since he believes that the criteria used to measure this concept are themselves closely correlated with race.

While the rabid extremes of the DOJ conduct in Miller should hardly stand as a minimum litmus test in such matters, it would seem to us that a plaintiff is required to do more than merely demonstrate that the DOJ had somewhat effective input in the process to trigger strict scrutiny. "It is not Justice Department interference per se that is the concern, but rather the fact that Justice Department pressure led the State to act based on an overriding concern with race." Abrams, 117 S. Ct. at 1934. And that is all the plaintiffs have really done here. The DOJ will almost always have this kind of input into the districting decisions of jurisdictions covered under section 5. At least in those jurisdictions, then, accepting the plaintiffs' minimal showing here would render Justice O'Connor's allowance for the creation of minority-majority districts of merely academic interest. In the process, such a rule would in practice narrow what is already a difficult passage through the Scylla of the Voting Rights Act and the Charybdis of Shaw nearly to the vanishing point--virtually every decision the jurisdiction made would run aground on the rocks of litigation. Here, for example, the City would have surely run into trouble had it disregarded the DOJ's suggestion regarding the creation of more minority concentrations. But by the mere fact of heeding these objections, plaintiffs claim, the City automatically exposed itself to a lawsuit triable under strict scrutiny. DOJ's overreaching will occasionally inflict this dilemma in any event--to exacerbate the problem by making it an ironclad rule would seem inadvisable.

Given the fact that the plaintiffs bore the burden of proof on this issue, and the presumption in favor of the Council's good faith, the plaintiffs needed to undercut the hypothesis that the City's plans were independently substantially justified by traditional districting factors. They failed to do so. An obvious mechanism would be to use circumstantial evidence based on the districts' shape and demographics to shore up the direct evidence of intent. But as discussed above, the plaintiffs' circumstantial evidence is inadequate to allow a finding that race predominated. Nor did the plaintiffs point to adequate direct evidence that the compliance with the DOJ's desires was the predominate, driving force behindthe plan, rather than merely one factor, along with traditional ones, in the overall equation. Cf., e.g., Abrams, 117 S. Ct. at 1934 (discussing direct evidence from drafter of plan that preceded the one challenged in Miller that he viewed draft including two majority-minority districts, that it was viewed as ridiculous and not making sense, and going on to note testimony of legislators that indicated that all other considerations yielded to DOJ desire to maximize minority districts).

And in any case, while the district court erred in categorically and totally dismissing evidence of intent garnered from prior plans, it was correct to point out that the state of mind involved in the prior plans is not of itself what is precisely and directly the ultimate issue before the Court in this case. We have noted in a different context that while under Hunter the discriminatory intent of the original drafter may carry forward despite subsequent judicial invalidation of the most obviously discriminatory provisions, intervening reenactment with meaningful alterations may render the current law valid. See Cotton v. Fordice, 157 F.3d 388, 391 (5th Cir. 1988) (while disenfranchisement constitutional provision originated in attempt to discriminate, subsequent reenactment with alterations approved by voters added categories of crimes originally excluded because they were not considered "Black" crimes and subtracted a less serious offense that had been considered a "Black" crime). We do not suggest that the changes implemented in 1993, 1995, and 1997 to the original framework were as dramatic as those in Cotton. However, that case broadly stands for the important point that when a plan is reenacted--as opposed to merely remaining on the books like the provision in Hunter--the state of mind of the reenacting body must also be considered.

Here, the passage of time lent an additional element to the Council's intent. Maintenance of established district lines is itself a traditional districting principle, and the City included it in its guidelines for the 1997 revision. And the adopted plan minimized the changes in district boundaries. Standing alone, that would not matter much here--the mere passage of six years cannot have hallowed the current districts with much familiarity or tradition. But in the context of the approaching census, this principle would seem to have some validity. As discussed above, the City had before it when it voted on the plan information indicating that a radical revision of its boundaries (for its soon to follow elections) will be necessary in light of the 2000 census. What the district court characterized as a "scorched-earth" redrawing of the districts--what Dr. Weber's views would have logically required to purge all taint of DOJ interference-would have little to recommend it under the circumstances. Such ground-up revisions are difficult to implement, and as one Council member argued, can be "wholly disruptive toward the political and neighborhood relationships we have established in the city." This fact hardly justifies the "freezing" of district lines in the manner the district court seemed to endorse. However, it does caution against wholesale alteration based on out-of-date census figures when the process will in any case have to be done in the immediate future. In the context of close proximity to the census (with new elections soon to follow), we think that residual doubt as to the appropriateness of summary judgment is extinguished by the fact that the City had valid reasons to adhere to its prior borders distinct from the grounds which led to their adoption.

Thus, although the issue on appeal is extremely close, we ultimately conclude that the district court correctly dismissed the plaintiffs' Shaw claim. On this record, we cannot say that plaintiffs have carried their burden and made a sufficient showing, either circumstantially or directly, to sustain a finding that race predominated in the districting. We stress that our decision here rests largely on the evidentiary shortcomings in the plaintiffs' case. We would also add that in our analysis of thedirect intent evidence we relied on the rapid approach of the census, and the City's expectation that a major overhaul of districts (and soon to follow elections) will occur at that time. Our opinion here cannot be read as a conclusive endorsement of these districts' constitutional status, since on better showing by the plaintiffs the result might well have been different. Nor can it be viewed as an open invitation to the City to maintain the current boundaries and overall structure in the redistricting that follows the 2000 census. See Campos v. City of Houston, 113 F.3d 544, 548 (5th Cir. 1997) (clarifying proper standard of measurement under section 2); id. at 545 (discussing pattern of election results); United States v. City of Houston, 800 F. Supp. 504, 508 (S.D. Tex. 1992) (Jones, J.) (discussing election results and question of safe versus marginal districts in the retrogression context).

Plaintiffs also contend that the votes of residents of several districts were devalued by the City's deliberate undersizing of minority voting districts. The heart of this one-person, one-vote claim is that the City, despite being aware that it contained pockets with extremely high ratios of noncitizens, improperly crafted its districts to equalize total population rather than citizen voting age population (CVAP). The plaintiffs argue that when total population is obviously an imperfect proxy for potential voters, the City is constitutionally required to use a more accurate measurement of voters. And using CVAP figures, it is clear that several Houston districts fall outside the ten percent threshold established as a safe-harbor for population variance in municipal election districts. The district court rejected this claim, and for slightly different reasons we agree.

The Equal Protection Clause requires that representatives to an elected body be drawn from voting districts of substantially equal population. Reynolds v. Sims, 84 S. Ct. 1362 (1964).14 This guarantee extends to local elections. See Avery v. Midland County, 88 S. Ct. 1114 (1968). However, because state and local elections are not subject to the explicit commands of the Constitution regarding federal elections, they are not held to a standard of absolute population equity. See Brown v. Thompson, 103 S. Ct. 2690, 2695-96 (1983) (state must make only a good faith effort to ensure population equity and may balance population against other legitimate concerns). If the maximum variation between districts exceeds a certain threshold, the state will be required to justify the variance by invoking such concerns. Mahan v. Howell, 93 S. Ct. 987 (1973) (16.4% variation justified by constraints of political subdivision lines). However, below a certain threshold the plaintiff has failed to establish a prima facie case and the districting body will not be required to justify minor variations. See Gaffney v. Cummings, 93 S. Ct. 2321, 2330 (1973). The Court has indicated that this threshold is ten percent. See, e.g., Connor v. Finch, 97 S. Ct. 1828, 1833 (1977) (setting threshold at ten percent). Here, the maximum variation between the City's districts was found by the district court to be 8.63% when measured by total population, and the plaintiffs have never claimed that using this population base the variation exceeds 10%. The plaintiffs wouldthus appear to have failed to make out a prima facie case.15

The plaintiffs argue, however, that the City's districts do shatter the ten percent threshold once the proper measurement of voters in utilized. They argue that they have thus made out a prima facie case of vote dilution. Plaintiffs contend that data available to the City indicated that areas with concentrated Hispanic populations had an extremely high number of noncitizens. They argue that given this well-known fact, the City should have recognized that total population would not serve as a meaningful proxy for potentially eligible voters--areas with concentrations of Hispanics would have a far larger population than potentially eligible voters. By nevertheless continuing to use total population, the City devalued the votes of residents in districts that did not contain concentrations of Hispanic voters. They argue that when measured against the proper standard, the City's districts displayed a maximum variance well in excess of the allowable ten percent. We have had cause before to note the high number of Hispanic residents of Houston who are not citizens. See Campos v. City of Houston, 113 F.3d 544, 547 (5th Cir. 1997) (census figures indicate 45.8% of adult Hispanics in City are noncitizens). And the plaintiffs have produced data that appears to clearly show that when the CVAP of Houston districts is compared, the maximum variance indeed exceeds the ten percent threshold. If the City were required by the Equal Protection Clause to use CVAP rather than total population in crafting its districts, then, summary judgment for the city would have been inappropriate. However, while this is a close question, we find that the choice of population figures is a choice left to the political process. Houston did not violate the Equal Protection Clause.

Plaintiffs' argument is straightforward. A Reynolds claim is, after all, a "-one person, one vote'" claim. See Reynolds, 84 S. Ct. at 1380 (quoting Gray v. Sanders, 83 S. Ct. 801, 809 (1963)). Since the inquiry focuses on the dilution of votes, it would be improper to allow the votes of two adult citizens to be weighed equally with the vote of a single adult citizen merely because the latter happened to live in proximity to a noncitizen ineligible to vote.16 Plaintiffs concede that undermany, if not most, circumstances this distinction will be irrelevant--generally, the ineligible to vote or to register to do so (whether felons, minors, or noncitizens) can be assumed to be evenly distributed throughout the area to be districted, and the usage of total population is thus an acceptable surrogate for measuring potential voters. When, however, a districting body knows that large numbers of those ineligible to vote are disproportionately concentrated in certain areas, it can no longer in good faith use total population as a proxy for potential voters. Instead, it is obligated to deploy a more sophisticated measurement, such as CVAP.

Two circuit courts have addressed this issue. The Ninth Circuit has found that total population is not only a permissible method to measure population when known significant concentrations of those not eligible to vote exist, but has also suggested that its usage may be required under the Equal Protection Clause in some circumstances. See Garza v. County of Los Angeles, 918 F.2d 763, 775 (9th Cir. 1990), cert denied, 111 S. Ct. 681 (1991) (approving court ordered plan using total population and opining that use of CVAP "would constitute a denial of equal protection to these Hispanic plaintiffs"). The Fourth Circuit, confronting the analogous issue of districting when persons below the voting age were unevenly distributed, has stated that the choice between total population or a measurement of potential voters is left to the legislative body. In doing so, it implicitly rejected the Ninth Circuit's contention that a districting body might be required by the Equal Protection Clause to use total population as a baseline. See Daly v. Hunt, 93 F.3d 1212, 1227 (4th Cir. 1996) (remanding on other grounds). However, in a powerful dissent in Garza, Judge Kozinski concluded that, at least when the plan is court ordered, the Constitution requires the use of a measurement that accurately tracks potential voters. See Garza, 918 F.2d at 785.

The Supreme Court has from the beginning of this line of cases been somewhat evasive in regard to which population must be equalized. See Reynolds, 84 S. Ct. at 1390 (state should attempt to apportion "so that each [districting unit] has an identical number of residents, or citizens, or voters"). However, it has also indicated with some clarity that the choice has political overtones that caution against judicial intrusion. See Burns v. Richardson, 86 S. Ct. 1286, 1297 (1966) (state not required to use total population figures, since the choice of measurement "involves choices about the nature of representation with which we havebeen shown no constitutionally founded reason to interfere"). The debate does indeed involve questions about the nature of representative democracy. As Judge Kozinski's dissent recognizes, and the Daly opinion outlines, there is nothing inherent in the concept of representative democracy that requires the embrace of the plaintiffs' position.

The choice presented is, in Judge Kozinski's terms, one between electoral equality and representational equality. While for the most part obscured by the happy coincidence that eligible voters will frequently track the total population evenly--in which case either measurement would produce similar results, and total population will probably be used since it is easier to gather--this is a fundamental and difficult question. If total population figures are used in an area in which potentially eligible voters are unevenly distributed, the result will necessarily devalue the votes of individuals in the area with a higher percentage of potentially eligible voters. The large populations in the other district will leverage the votes of the smaller number of eligible voters there. However, if a more carefully calibrated measurement is embraced in order to serve this principle of individual electoral equality, the area with the smaller number of voters will find itself relatively disadvantaged. Despite the factthat it has a larger population--and thus perhaps a greater need for government services than the other community--it will find that its political power does not adequately reflect its size. In addition, it could be argued that because the representative chosen from such a district will have a larger number of constituents, the ability of her constituents--whether or not they are potential voters--to petition and voice their opinions will be proportionately reduced. See Garza, 918 F.2d at 775. But see Daly, 93 F.3d at 1226-27 & n.12 (expressing doubt regarding validity of equal access argument). If one accepts the principle of representational equality--that representatives are chosen by a district's voters, but should represent all persons resident therein--these results may be unacceptable. The choice between these two models is stark, and because it is rarely encountered, historical and legal guidance is sparse.

Judge Kozinski makes a powerful case that the general tenor of the Court's opinions mandates protection of the individual potential voter. Certainly, to take just a few examples contained in Judge Kozinski's comprehensive summary, there is ample language in the opinions that strongly implies that it is the right of the individual potential voter that must be protected. See, e.g., Reynolds, 84 S. Ct. at 1384 ("the Equal Protection Clause guarantees the opportunity for equal participation by all voters in the election of state legislators. Diluting the weight of votes because of place of residence impairs basic constitutional rights under the Fourteenth Amendment . . . . To the extent that a citizen's right to vote is debased, he is that much less a citizen . . . . 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"); Gaffney, 93 S. Ct. at 2328 (noting that among the variations that made absolute fidelity to population equality impossible was the gap between total population and potential voters, and stating that "if it is the weight of a person's vote that matters, total population...may not accurately reflect that body of voters").

But as Judge Kozinski admits, and as the Daly court highlights, other language can be found that implies that representational equality is the ideal. See, e.g., Mahan, 93 S. Ct. at 983 ("the basic constitutional principle [is] equality of population among the districts"). While it does appear that the numerical weight of references is on the side of electoral equality, it is difficult to attach controlling significance to this fact, since in almost all cases the Court was dealing with situations in which total population was presumptively an acceptable proxy for potentially eligible voters.Under such circumstances, we would expect to find the terms used interchangeably, with perhaps a slight bias toward the more historically resonant phrase--unquestionably, one-person, one-vote. Support for this interpretation is provided by the fact that some of the language Judge Kozinski relies on is contradicted by statements within the same opinion. See Reynolds, 84 S. Ct. at 1385 ("We hold that, as a basic constitutional standard, the Equal Protection Clause requires that the seats . . . must be apportioned on a population basis"); Westberry v. Sanders, 84 S. Ct. 526, 530 (1964) ("To say that a vote is worth more in one district than in another would . . . run counter to our fundamental ideas of democratic government"); id. at 535 ("our Constitution's plain objective of making equal representation for equal numbers of people the fundamental goal for the House of Representatives"). While fully recognizing the force of Judge Kozinski's argument, we think that the cases he cites are not decisive, given the strong possibility that much of the language he relies on may be traceable to the use of terms interchangeably in a context where their meaning does not diverge.

Instead, we turn to Burns, where the Court directly confronted an actual differential between the concepts. We read Burns as compelling rejection of the Garza majority view that the Equal Protection Clause mandates inclusion of aliens in the population base of electoral districts against which the equality requirements of Reynolds are applied. Burns states:

"Neither in Reynolds v. Sims nor in any other decision has this court suggested that the States are required to include aliens, transients, short-term or temporary residents, or persons denied the vote for conviction of crime in the apportionment base by which their legislators are distributed and against which compliance with the Equal Protection Clause is to be measured. The decision to include or exclude any such group involves choices about the nature of representation with which we have been shown no constitutionally founded reason to interfere. Unless a choice is one the constitution forbids, [citation] the resulting apportionment base offends no constitutional bar, and compliance with the rule established in Reynolds v. Sims is to be measured thereby.

. . . .

Hawaii's special population problems might well have led it to conclude that state citizen population rather than total population should be the basis for comparison." Id. at 1296-97 (footnote omitted).

For this reason, we believe the district court erred to the extent it relied on the Garza majority dicta.17 Judge Kozinski argues that Burns should be determinative for electoral equality, since there the Court allowed a plan that used an even more nuanced measure of voting strength--registered voters. But we are reluctant to read Burns' allowance of such a measure into a command in the face of what appears to us to be a clear statement to the contrary in the same opinion--that the choice between measurements "involveschoices about the nature of representation with which we have been shown no constitutionally founded reason to interfere". Burns at 1297.

We also note that the drafters of the Fourteenth Amendment, on which Reynolds itself rests, do appear to have debated this question, and rejected a proposal rooted in--among other things--the principle of electoral equality. On December 5, 1865, Representative Stevens introduced a constitutional amendment which apportioned congressional representation among the states "according to their respective legal voters; and for this purpose none shall be named as legal voters who are not either natural-born citizens or naturalized foreigners." See Joseph T. Sneed III, Footprints on the Rocks of The Mountain: An Account of the Fourteenth Amendment at 35 (1997) (Footprints). Debates over the precise basis for apportionment of Congress among the states proved a contentious issue throughout the process that led to the creation of section 2 of the Amendment. The overriding question driving this debate was the issue of how to deal with the denial of the franchise to African-Americans. However, as the inclusion of a citizenship provision indicates, an undercurrent of this debate was the recognition by many representatives that aliens were unevenly distributed throughout the country. In addition, several western states contained an overabundance of males, which, as women were generally not eligible to vote, would disadvantage them relative to settled eastern areas in which the genders were balanced if total population rather than eligible voters were used. Some of the debate over whether to base apportionment on potential voters, citizens, or population can be tied to this issue.18

To be sure, the overall context in which the amendment was drafted prevents any firm conclusion being drawn as to the framer's intent regarding the question before us. However, we find it is of some significance that the proponents of the Fourteenth Amendment had a meaningful debate on the question, which cannot be said to have been definitively resolved.19 While the final version of section 2 provided generally for the use of total population figures for purposes of allocating the federal House of Representatives among the states, it also included a mechanism to insure that egregious departures from the principle of electoral equality--the disenfranchisement of adult male "citizens"--would be penalized. Bearing in mind that analogies drawn from the federal system are not always applicable to the states, see Reynolds, 84 S. Ct. at 1386-89, we have some difficulty in reading the Equal Protection Clause to require the adoption of a particular theory of political equality.20

While hardly determinative, our review of the history of the amendment cautions against judicial intrusion in this sphere--either for or against either particular theory of political equality. Given this, and given the Court's failure, on our reading, to speak clearly to such a vital question, we see no justification to depart from the position of Daly. We reject the conclusions of both the dissent in Garza and any reading of the majority opinion in that case that would mandate the use of total population figures on equal protection grounds. We thus conclude that the district court correctly determined that the plaintiffs failed to create an issue of genuine material fact in respect to the City's adherence to the one-person, one-vote principle.

The record is not such as to sustain a finding that race predominated in the City's decision-making. While the plaintiffs were able to point to bits and pieces of the City's plan that might have created a valid Shaw claim, they failed to produce evidence that would have surmounted the summary judgment threshold. The propriety under the Equal Protection Clause of using total population rather than a measure of potential voters also presents a close question. But in face of the lack of more definitive guidance from the Supreme Court, we conclude that this eminently political question has been left to the political process.

For the reasons stated, the judgement of the district court is

AFFIRMED.