PER CURIAM. In the spring of 1991, the Census Bureau furnished the State of Wisconsin with a detailed breakdown of the results of the 1990 decennial census. The breakdown showed that as a result of population shifts since the 1980 decennial census, the Wisconsin legislature was malapportioned--the shifts had produced large discrepancies in population between districts. The shifts had probably caused a violation of the Voting Rights Act, 42 U.S.C. § 1973, as well, because they had resulted in blacks' being "packed" into districts in Milwaukee, thus "wasting" black votes and therefore, arguably, denying blacks the reasonable opportunity to select legislators of their choice that the Act guarantees them. Both houses of the Wisconsin legislature have a Democratic majority, but not a large enough one to override vetoes by the state's Republican governor. For that or other reasons, no bill to reapportion the legislature had been enacted into law when, on January 30 of this year, several Republican legislators filed this suit challenging the current apportionment of the legislature as unconstitutional and violative of the Voting Rights Act. Davis v. Bandemer, 478 U.S. 109, 92 L. Ed. 2d 85, 106 S. Ct. 2797 (1986). This three-judge district court was convened pursuant to 28 U.S.C. § 2284. The Democratic leaders of the Wisconsin legislature were permitted to intervene, as were a number of groups, including the Wisconsin Education Association Council, and individuals, including Annette Williams, a black representative from Milwaukee, and several other black and Hispanic legislators. The case was expedited to enable the state primary and general elections to proceed on schedule in the new districts. Shortly before the evidentiary hearing that we had scheduled for the week of April 27, the legislature passed a reapportionment bill, which the governor vetoed. The hearing, held in Madison on April 27 and 28, focused on four out of the ten plans that had been submitted. (The reason for the selection of those plans is discussed later.) Expert evidence in support of the various plans was introduced in written form, so that the hearing could be devoted to cross-examination of the experts and to opening and closing arguments of counsel.
We have now arrived at a decision and this opinion sets forth its legal and factual basis. Fed. R. Civ. P. 52(a). We discuss the malapportionment issues first and the Voting Rights Act second, but preface our discussion with a brief description of the political and demographic character of Wisconsin.
The state is large but thinly populated, with a shade under 5 million people, of whom about a fifth live in Milwaukee County in the southeastern corner of the state. The state is largely white, the 5 percent that is black being concentrated in the city of Milwaukee. There are also small Hispanic, Asian, and American Indian minorities. The state has a tradition of political independence, and although the balance in recent years has tipped slightly in favor of the Democrats, popular Republicans such as Reagan and Thompson (the present governor) have carried the state--in Thompson's case, by a lopsided margin. State law fosters (or perhaps reflects) political independence by allowing any eligible voter, regardless of political affiliation, to vote in either primary. A high voter turnout in general elections is facilitated by allowing voters to register at the same time that they vote.
The Census Bureau, for its purposes, divides the state into thousands of census blocks, the population of which varies from 0 to 3,000 people. The political subdivisions of the states include the usual--counties, towns, etc.--as well as legislative districts. The smallest subdivision is the ward. Although state law requires a ward to have at least 300 residents, there are exceptions, and some wards, we were told, have as few as 6 people in them. The entire state is divided into wards, and all wards are nested within the larger subdivisions; that is, no ward is in two counties, two towns, two assembly districts, etc. However, a number of wards "split" census blocks; that is, the block may be part in one ward and part in another. The state has 99 assembly districts, with an average of some 49,000 people per district. There are 33 senatorial districts, each composed of three assembly districts. In part to achieve population equality some districts split other political subdivisions (other than wards)--some for example cross county or town lines.
Legislative districting has generated political controversy at least since England's "rotten boroughs." The reason is that any disparity in the number of voters in different districts dilutes the influence of some voters on the composition of the legislature. If for example one district has ten times as many electors as another, the electoral influence of each voter in the first district will be one-tenth that of each voter in the second district. Inequality in voting power as a consequence of disparities in population among legislative districts has been the particular target of the Supreme Court's reapportionment decisions, which have decreed the norm of "one person, one vote." Reynolds v. Sims, 377 U.S. 533, 577, 12 L. Ed. 2d 506, 84 S. Ct. 1362 (1963). The norm, however, is unattainable for two reasons: measurement error, and the presence of competing norms that cannot be ignored. The decennial census involves both undercounting and overcounting, and the errors are not completely random and therefore do not cancel out. Tucker v. Department of Commerce, 958 F.2d 1411, 1412-13 (7th Cir. 1992). Moreover, population shifts occur in the interval (here two years) between the census and the reapportionment based on it. As for competing norms: there is a nearly infinite set of district configurations that would generate approximate population equality across districts, and no one supposes that a court should be indifferent among all members of the set. It would be possible to create a district of 49,000 Wisconsinites by assembling census blocks from all over the state, by joining a Milwaukee neighborhood with a rural area in the northwestern corner of the state, hundreds of miles away, by cutting a corridor 200 hundred miles long and a quarter of a mile wide that would snake through the state, and in a million other ways. It would be possible to create a senatorial district by combining three widely separated assembly districts. With the right computer program a complete reapportionment map for the state can be created in days and modified in hours and we have no doubt that the parties examined hundreds of possible plans before submitting the handful that we have been asked to consider.
The objections to bizarre-looking reapportionment maps are not aesthetic (except for those who prefer Mondrian to Pollock). They are based on a recognition that representative democracy cannot be achieved merely by assuring population equality across districts. To be an effective representative, a legislator must represent a district that has a reasonable homogeneity of needs and interests; otherwise the policies he supports will not represent the preferences of most of his constituents. There is some although of course not a complete correlation between geographical propinquity and community of interests, and therefore compactness and contiguity are desirable features in a redistricting plan. Compactness and contiguity also reduce travel time and costs, and therefore make it easier for candidates for the legislature to campaign for office and once elected to maintain close and continuing contact with the people they represent. Viewing legislators as agents and the electorate as their principal, we can see that compactness and contiguity reduce the "agency costs" of representative democracy. But only up to a point, for the achievement of perfect contiguity and compactness would imply ruthless disregard for other elements of homogeneity; would require breaking up counties, towns, villages, wards, even neighborhoods. If compactness and contiguity are proxies for homogeneity of political interests, so is making district boundaries follow (so far as possible) rather than cross the boundaries of the other political subdivisions in the state.
Speaking of disenfanchisement and gerrymandering (in its broadest sense of partisan districting), redistricting a legislative body whose members have staggered terms unavoidably creates the former, though only temporarily, and with it possibilities for the latter. When a senatorial district line is redrawn, some persons who formerly resided in an even-numbered district, and hence last voted for senator in 1988 and would, but for redistricting, have voted for senator again this year, now find themselves in an odd-numbered district and so must wait until 1994 to vote for senator again. In effect they are "disenfranchised" from one election. (An equal number, however, get to vote in one more election--they are allowed to vote twice for senator in four years, 1990 and 1992. But taking away one person's vote is not remedied by giving someone else two votes.) The effect is not limited to voters. Suppose that you are a senator representing an even-numbered district. Therefore your term expires at the end of this year. Suppose that as a result of redistricting you find yourself a resident of an odd-numbered district. You cannot run in that district this year, because odd-numbered districts do not vote for senator again until 1994. You would have to move to an even-numbered district and run against the incumbent, assuming he was seeking reelection.
The broader problem is that if as a result of redistricting two incumbents find themselves residents of the same district, and neither decides to retire or to move to another district, they must run for "reelection" without any of the usual advantages of incumbency, because their opponent is also an incumbent. In recent years (though public opinion polls suggest that this year may be different), incumbents have had a marked advantage in electoral contests with newcomers, so a partisan redistricting plan will seek to "pair" (place in the same district) as many legislators of the opposite party, and as few of their own party, as possible; more precisely, pair as many opposing (and as few of one's own) legislators who plan not to retire, or to move their legal residence to another district, as possible.
With exact population equality unattainable and in any event not the only goal of redistricting, it is apparent that the design of a reapportionment plan that will be "best" in terms of the goals of representative democracy is a daunting task, especially for judges. Although two of the judges of this court are long-term residents of Wisconsin, none of us is familiar with every part of the state or with the intricacies of local government, and the parties have not supplied us with enough information to enable us to become expert reapportioners. They could not have done this if they had wanted to, because the speed with which this litigation has had to be conducted in order to enable the fall primary and general elections to proceed in orderly fashion has necessarily limited the scope of inquiry.
Our task would be easier if we were reviewing an enacted districting plan rather than being asked to promulgate one ourselves. If for example the bill containing the legislative plan had been signed by Governor Thompson, and was being challenged as unconstitutional, our task would be to decide not whether the plan was the best possible but whether it struck a reasonable balance among the considerations enumerated above, starting with approximate numerical equality among districts. The legislative plan never became law, however. The only plan that became law, and therefore the only plan that is challenged, is the existing apportionment of the Wisconsin legislature, which all parties concede is unconstitutional. The issue for us is therefore remedy: not, Is some enacted plan constitutional? But, What plan shall we as a court of equity promulgate in order to rectify the admitted constitutional violation? What is the best plan?
Mindful of our limitations, we asked the parties at the outset whether they had any objection to our treating their plans in the manner of "final offer arbitration," that is, to our selecting the best of the submitted plans rather than trying to create our own plan, whether from the ground up or out of bits and pieces of the plans submitted by the parties. Only the Wisconsin Education Association Council objected, and it did so weakly. We permitted the parties to submit multiple plans and to amend their plans. Although a total of ten plans were submitted, the focus of our deliberations has been on four plans--Prosser IA and IIIA, which are the Republican plans; the legislative plan, which is the Democratic plan; and Representative Williams's plan. The other plans that like Williams's plan would create four rather than five black majority districts in Milwaukee (the George, Ortiz, and Berry plans) suffer from the same infirmities as her plan and need not be discussed separately. Nor the Coggs and WEAC plans, neither of which differs substantially from the legislative plan.
After considering the plans, we have decided to retract our threat to choose the "best" no matter how bad it was. The best plans were Prosser IIIA and the legislative plan, and both bear the marks of their partisan origins. We have decided to formulate our own plan, which combines the best features of the two best plans.
Prosser IA is defended on the ground that it achieves perfect numerical equality. It does this by building districts out of census blocks, without regard for wards (remember that many wards split census blocks). Prosser IIIA also achieves close to perfect numerical equality, but it adheres to ward boundaries and therefore splits some blocks. The plaintiffs argue that IA is better, because splitting blocks makes true numerical equality unattainable. The argument rests on the fallacy of delusive exactness. The decennial census is not accurate, and the 1990 decennial census is already out of date. When blocks are split in the forming of wards, the population of the ward is estimated by estimating the proportion of residents of each split block in the ward to which they are assigned. The estimation procedure is not arbitrary, but is based on counting dwellings and asking building managers how many people are in their building. Doubtless there are errors but so far as anyone has suggested to us they are random--which cannot be said of all the errors in the census count itself. As far as the trivial differences in population equality between Prosser IA and IIIA are concerned, they are entitled to no consideration. We ought to be realistic. If one assembly district has 49,000 residents, and another has 49,500, there is no basis for assuming any dilution of voting power in the second district, for not all residents of a district vote--not all are eligible to vote (children, for example, are not eligible). Even if it were assumed falsely that the ratio of voters to total population were the same in the two districts, the dilution in any voter's electoral power resulting from a 1 percent difference in number of voters would be too trivial to register in the most sensitive analysis of political power.
We can therefore narrow our focus still more by rejecting Prosser IA because it gratuitously breaks up wards. Wards are not sacred, but they are the basic unit of Wisconsin state government for voting purposes. You vote by ward. Under Prosser IA, people in the same ward would be voting in different races. This would not be a major inconvenience, but it would be some and there is, as we have pointed out, no offsetting gain in population equality; the apparent gain is a statistical illusion.
Between Prosser IIIA and the legislative plan, the differences are few. Both districting plans create districts having a high degree of compactness and contiguity, with one exception. Towns in Wisconsin are permitted to annex noncontiguous areas, and this is sometimes done. The legislative plan treats these "islands," as the noncontiguous annexed areas are called, as if they were contiguous, but the Prosser plans require literal contiguity and therefore always place the area between an island and the town that owns it in the same district with the town and the island. Since the distance between town and island is slight, we do not think the failure of the legislative plan to achieve literal contiguity a serious demerit; and we note that it has been the practice of the Wisconsin legislature to treat islands as contiguous with the cities or villages to which they belong. Wis. Stat. § § 4.001(3), 5.15(b). We are not persuaded by the plaintiffs' argument that the Wisconsin constitution requires literal contiguity.
Both Prosser IIIA and the legislative plan follow the boundaries of the other political subdivisions more or less, though they occasionally split counties, cities, and towns (but never wards). Neither party supplied enough information to enable us to determine whether the splits of the one are more serious than those of the other in terms of breaking up populations that are homogeneous in their need of or demand for governmental services, and other relevant criteria of community of interest. Prosser IIIA temporarily "disenfranchises" a considerably greater number of voters, largely because of its authors' decision to change an odd-numbered district in Milwaukee to an even-numbered district and to restore the balance by changing an even-numbered district elsewhere in the state to an odd-numbered district, thereby disenfranchising (in the 1992 election) the residents of the latter district. The purpose of the decision was to give minority residents of Milwaukee an earlier shot at another senate seat. Although "a temporary dilution of voting power that does not burden a particular group does not violate the equal protection clause," Republican Party v. Keisling, 959 F.2d 144, 145-46 (9th Cir. 1992) (per curiam), and is an inevitable concomitant of redistricting, it is not something to be encouraged. At the hearing the plaintiffs announced their readiness to abandon this feature of their plan, but they never submitted an amended plan.
With the plans similar in most respects, much of the hearing focused on the question of political fairness, or gerrymandering (broadly defined). The plaintiffs told us that political fairness is irrelevant--that their plan or plans should be preferred, regardless of political fairness, because their plans produce the more perfect numerical equality and that's all that counts. (Yet inconsistently they argue that their plans should also be preferred because they achieve greater contiguity.) But we have seen that there is no realistic basis for supposing that their plans produce greater equality, and if they do, the margin is too slight to matter. What is true is that if we were reviewing an enacted plan we would pay little heed to cries of gerrymandering, because every reapportionment plan has some political effect, and so could be denounced as "gerrymandering" committed by the party that had pressed for its enactment. Gaffney v. Cummings, 412 U.S. 735, 752-53, 37 L. Ed. 2d 298, 93 S. Ct. 2321 (1973); see also Davis v. Bandemer, supra; Peter H. Schuck, "The Thickest Thicket: Partisan Gerrymandering and Judicial Regulation of Politics," 87 Colum. L. Rev. 1325 (1987). But we are not reviewing an enacted plan. An enacted plan would have the virtue of political legitimacy. We are comparing submitted plans with a view to picking the one (or devising our own) most consistent with judicial neutrality. Judges should not select a plan that seeks partisan advantage--that seeks to change the ground rules so that one party can do better than it would do under a plan drawn up by persons having no political agenda--even if they would not be entitled to invalidate an enacted plan that did so.
Prosser IIIA, particularly on the senatorial side, is the more partisan, as it appears designed to decapitate the Democratic leadership in the senate. It does this by placing 4 of the senate's Democratic leaders (including both the majority leader and the assistant majority leader), all of whom represent even-numbered districts, in odd-numbered districts, thus preventing them from running for the senate this fall unless they move their legal residence to an even-numbered district, where they would have to run against an incumbent unless the incumbent decided to move or retire. (In a post-hearing submission, the plaintiffs offered to renumber the leaders' districts. The offer, unaccompanied as it is by any explanation of the effect of changing the numbers of other districts to permit the renumbering of the leaders' districts, comes too late.) On the assembly side, Prosser IIIA would pair 12 Democrats and 4 Republicans. Since there are more Democrats than Republicans in the assembly, a nonpartisan plan would presumably pair more Democrats than Republicans--but not this many more, because the percentage of Democrats in the assembly is 58 percent, not 75 percent. Of course the political impact of pairing depends on the plans of the incumbents. The plaintiffs argue that although the legislative plan appears to be less one-sided in favor of Democrats than Prosser IIIA appears to be one-sided in favor of Republicans, in fact the Democrats have been careful to pair only Democrat incumbents one of whom is planning to retire or move anyway. However, insufficient evidence concerning incumbents' plans, when those plans were made, and how firm they are was presented to enable us to go beyond the bare statistics of pairing.
In defense of the political fairness of Prosser IIIA the plaintiffs asked us to compare the results that it would produce with the results of two "base races." To explain, a politically fair apportionment plan is one that will produce a legislative composition that reflects the respective voting strengths of the parties in the state--but how is that strength to be gauged? Especially in Wisconsin, a state of ticket-splitters and independents, elections are influenced by a lot more than the candidates' party labels. Political scientists tell us, what is anyway common sense, that party labels are more likely to matter the more obscure the office to be filled by an election is, that is, the fewer other factors besides party identification are likely to be in play in the electoral contest. Such an election is a "base race" that can be used to infer the respective strength of the parties in the various parts of the state. The plaintiffs' experts selected as the base races for this case the last two elections (1986 and 1990) for state treasurer. They averaged together the votes for the respective parties in the two races and then determined how each party would have fared if the legislature were elected under the Prosser IIIA districting plan and the vote for each party's candidates were identical to the vote in each district for state treasurer in 1986 and 1990.
There was an intermediate step, however. Remember that in a districted legislature, as distinct from one in which legislators are chosen by proportional representation, small differences in voting strength can translate into large differences in representation (making it a little unclear what is meant by saying that a nonpartisan reapportionment will produce a legislature that "reflects" the respective voting strengths of the parties in the state). In the two "base races" selected by the plaintiffs the Republicans got only 48 percent of the vote. Were this their average vote in a legislative election they might get far fewer than 48 percent of the seats. So the experts added 2 percent to each district's Republican vote (in the 1986 and 1990 state treasurer's races), on the theory that if the Republicans in an election for state legislators held today had the same distribution of voting strength across districts that they had in the base races, except that they got an extra 2 percent in each district so that their average voting strength was 50 percent, then if this average strength translated into 50 percent or fewer Republican seats in the legislature the plan could not be pronounced politically unfair. And in fact Prosser IIIA would give the Republicans only 44 percent of the seats on these assumptions.
Unhappily for the plaintiffs, the ground for using the 1986 and 1990 state treasurer's races as base races was destroyed in cross-examination. The distinguished political scientist who conducted the base race analysis for the plaintiffs is not a Wisconsinite or familiar with Wisconsin politics, and he relied totally on the selection of base races by another expert, who while a reputable political scientist at the University of Wisconsin is also a high-level Republican activist. Cross-examination brought out that the state treasurer's race in Wisconsin; far from being a quiet arena for old-fashioned party politics, is riven by special factors. Oddest of all is the fact that, from time immemorial until 1990, the occupant of this office had always been named Smith. The latest Smith (a Democrat) had won reelection by a large margin in 1986. Four years later he stood for reelection once again, and this time, after charges of improprieties were leveled against him and widely publicized, was defeated by his Republican opponent. The victor was a woman and this may have played a role too. Whether party loyalty played any role in the election is doubtful and certainly unproven.
The alternative to finding a base race somehow purged of nonpartisan considerations--a snipe hunt, in all likelihood--is to average the results of a number of elections, on the theory that nonpartisan factors will tend to be randomly distributed and therefore will cancel out, leaving party loyalty as the only factor differentiating the vote totals of the two parties. (A possible adjustment would be to discount each election result by 1 minus the percentage turnout, on the theory that nonpartisan factors tend to increase the turnout. That was not done here.) The Democrats tendered the results of 11 state-wide races, going back to 1982, the composite results of which establish the fairness of the legislative plan in just the same way as the plaintiffs' analysis establishes the fairness of their plan. The plaintiffs argue that 1982 is too long ago, and they have a point, and that 11 is a small sample (though larger than their own sample of 2), and this is also a point. The Democratic sample mysteriously omits several races, and since the district totals of those races are not in the computer that the parties to this litigation have been sharing and cannot be placed in the computer within the time constraints of this litigation, we cannot construct a more comprehensive sample. We conclude that whatever merits base race analysis may have when the data are adequate to conduct such an analysis, they are not adequate here.
Our evaluation of the competing plans must remain tentative until we consider the voting Rights issue, and let us turn to it now. The voting Rights Act authorizes and in some instances compels racial gerrymandering in favor of blacks and other minorities. Because the Act implements the Fifteenth Amendment, it is constitutional despite its discriminatory character. United Jewish Organization v. Carey, 430 U.S. 144, 159, 51 L. Ed. 2d 229, 97 S. Ct. 996 (1976). The wisdom and seemliness of the Act have been questioned, even by radical blacks. Lani Guinier, "The Triumph of Tokenism: The voting Rights Act and the Theory of Black Electoral Success," 89 Mich. L. Rev. 1077 (1991). Because blacks vote heavily Democratic, majority or supermajority black districts produce lopsided Democratic majorities, resulting in a wastage of Democratic votes and a weakening of the Democratic party--the party of most blacks. But none of that is our business.
From our earlier discussion it might seem that the Act's objectives would be best achieved by creating districts in which blacks were 51 percent of the population, since additional black voters would be wasted and would be better off being shifted to districts in which their votes might affect the outcome. But because a disproportionate number of blacks are below voting age, and because turnout among blacks is generally much lower than among whites, a 1 percent margin in population would not translate into a 1 percent margin in voting strength A rule of thumb has emerged in the cases that to give blacks a reasonable assurance of obtaining a majority of votes in a district the population of the district must be at least 65 percent black (50 percent plus 5 percent to reflect the lower average age of blacks and hence lower voting population, 5 percent to reflect a lower fraction of registered voters, and 5 percent to reflect a lower turnout) and the voting population at least 60 percent black. Ketchum v. Byrne, supra, 740 F.2d 1398, 1415-16 (7th Cir. 1984). We do not understand the separate criteria for overall population and voting-age population. Population is relevant in a districting case only as a proxy for voting population, McNeil v. Springfield Park District, 851 F.2d 937, 944-45 (7th Cir. 1988), so if one has the latter figure, as we do, one does not need the former. Ketchum v. Bryne, supra, 740 F.2d at 1413. We shall therefore confine ourselves to the 60 percent voting population (i.e., voting-age population) criterion.
We said at least 60 percent but we could equally well have said at most. For votes above the level needed to elect a representative of one's choice are wasted and had best, as we have said, be reallocated to another district, where they might make a difference. Suppose the choice were between the following pairs of districts: one pair consisting of one district that was 60 percent black and one that was 40 percent black, and the other pair consisting of one district that was 100 percent black and one that was zero percent black. Under the first choice, blacks could elect one representative of their choice and influence the election of one other representative. Under the second choice, they could elect one black representative and have no influence on the election of another representative. So the first would be superior in terms of the goals of the Voting Rights Act. Indeed, it is because, as a result of population shifts, the present 3 predominantly black assembly districts (and 1 black senatorial district) in Milwaukee have black voting populations far in excess of 60 percent that the present districting violates the Act.
Both Prosser IIIA and the legislative plan propose to create 5 black majority assembly districts in Milwaukee. In Prosser IIIA, 4 of these districts would have voting populations 60-61 percent black and the other 58.34 percent. The legislative plan would yield 3 districts with a black voting population of 60-61 percent, and the other 2 would have 59.78 and 59.87 percent respectively. These of course are small differences. The only substantial differences between the plans are two: the legislative plan would pair more incumbents, and at the same time would include more blacks in the 5 majority districts plus 1 more district, an "influence" district, that is, a district in which blacks comprise a sufficiently large fraction of the voting population to constitute an effective interest group, though not one with majority control. Both plans contain a black "influence" assembly district but the legislative one creates a larger black bloc in that district--24 percent versus 18 percent. The pairing of incumbents has no relation that we can see to the purposes of the Voting Rights Act. The creation of a stronger "influence" district, however, is a modest plus from the Act's standpoint; the skepticism expressed in McNeil v. Springfield Park District, supra, 851 F.2d at 947, on this score concerned the distinct issue whether failure to create such a district could be an actual violation of the Voting Rights Act. Both plans--indeed all plans, including our own--create a Hispanic "influence" assembly district in Milwaukee. The Prosser plans split Indian reservations; the legislative plan does not.
The real challenge on the Voting Rights Act front comes from Representative Williams. She (along with Representative Coggs and Senator George) wants 4 black majority districts instead of 5 because she fears that 60 percent really is not enough to guarantee black voting control of a district. She has a point. Even if all blacks vote for black candidates, if two blacks and one white run in the Democratic primary in a district in which blacks cast 60 percent of the votes, the black candidates may split the black vote and the white candidate win. (That happened, with the colors reversed, in a recent mayoralty election in Chicago.) Then too the 60 percent rule of thumb may not give adequate weight to the very low turnout of black voters in Milwaukee. Williams presented evidence that the last time she ran, even though the population of her district is 74 percent black, the number of votes cast by whites was only 100 fewer than the number cast by blacks. But she ran unopposed, and her expert witness conceded that, if she had had a white (or perhaps any) opponent, black voters would have turned out in greater number.
From the standpoint of a black incumbent concerned only with his or her reelection prospects in a potential contest with a white, the more blacks in the district the better. Williams's expert witness testified that to be safe for blacks a district must have a population that is 80 percent black. But an incumbent's standpoint is too limited a one. The goal of the Voting Rights Act is to enhance the electoral power of minority voters, rather than to maximize the electoral prospects of minority incumbents. This goal implies a tradeoff between the size of the minority supermajority in a minority district and the number of such districts. Under Williams's plan, black incumbents' prospects would be maximized but at the expense of the number of minority districts. The case against the 60 percent rule of thumb has not been made. The 60 percent rule provides a particularly comfortable margin here because of Wisconsin's "same day" registration rule. A person can register to vote at the polls on election day simply by producing an envelope with his name and address on it as proof of residence in the ward in which he is voting. As a result of this system, a very high percentage of registered voters actually vote, and the percentage of registered black voters who vote appears to be the same as that of white voters.
Hints have been dropped in newspaper articles that the legislative plan is designed to "get" Williams for breaking with the Democratic leadership on key issues. While no effort was made to prove this, and its relation to the purposes of the Voting Rights Act is in any event obscure, we are puzzled by the fact that the legislative plan pairs all the black incumbents in Milwaukee.
Our plan, what we call "the court plan," avoids this and other problems with the parties' plans. The court plan, set forth in the judgment order that accompanies this opinion, is based on the two best submitted plans--Prosser IIIA and the legislative plan. It preserves their strengths, primarily population equality and contiguity and compactness, and avoids their weaknesses. No useful purpose would be served by describing the boundaries of the 99 assembly and 33 senatorial districts created by the plan, but we shall compare its salient characteristics with those of the plans on which we drew. The court plan's total deviation from exact population equality is .52 percent, compared to .15 for Prosser IIIA and .52 for the legislative plan, and its mean deviation is .10 percent, compared to .03 for Prosser IIIA and .11 for the legislative plan; all these figures are well within the 1 percent margin of error. The court plan creates a black senatorial district in Milwaukee; the black voting-age population of the district is 59.8 percent, which is essentially the same as in Prosser IIIA and the legislative plan. The black "influence" senatorial district has a black voting-age population of 45 percent in our plan, which is the same as Prosser IIIA and slightly lower than the legislative plan. It creates five black-majority assembly districts, and one black-influence assembly district, having black voting-age populations essentially identical to those in Prosser IIIA. In number of splits the court plan falls in between the legislative plan and Prosser IIIA; it splits 115 political subdivisions smaller than counties, compared to 108 for the legislative plan and 130 for Prosser IIIA. It temporarily "disenfranchises" 257,000 voters compared to 200,000 for the legislative plan and 392,000 for Prosser IIIA. The court plan splits no Indian reservations.
Finally, the court plan, far as we are able to judge, creates the least perturbation in the political balance of the state. We have explained why the parties' submissions do not permit meaningful bace-race comparisons, but such as they are, these comparisons suggest that the court plan is the least partisan. More important, the court plan pairs only 16 incumbents in both houses of the legislature, and only 6 of the same party (4 Democrats and 2 Republicans). Prosser IIIA, in contrast, pairs 35 incumbents, including 22 of the same party: 18 Democrats and 4 Republicans. The legislative plan pairs 31 incumbents, 22 of the same party: 12 Democrats and 10 Republicans. It is possible to object that in protecting incumbents, our plan perpetuates and entrenches political imbalances created by the existing, and unconstitutional, apportionment. But while the existing apportionment is conceded to be unconstitutional as a consequence of population shifts, the plaintiffs have not shown that it was politically biased from the start or that their own plan corrects such political bias as there may have been-without overcorrection.
IT IS ORDERED that effective June 2, 1992, the 99 assembly districts described in part II of this order are organized into 33 senate districts as follows:
The combination of the 1st, 2nd and 3rd assembly districts.
The combination of the 4th, 5th, and 6th assembly districts.
The combination of the 7th, 8th, and 9th assembly districts.
The combination of the 10th, 11th, and 12th assembly districts.
The combination of the 13th, 14th, and 15th assembly districts.
The combination of the 16th, 17th, and 18th assembly districts.
The combination of the 19th, 20th, and 21st assembly districts.
The combination of the 22nd, 23rd, and 24th assembly districts.
The combination of the 25th, 26th, and 27th assembly districts.
The combination of the 28th, 29th, and 30th assembly districts.
The combination of the 31st, 32nd, and 33rd assembly districts.
The combination of the 34th, 35th, and 36th assembly districts.
The combination of the 37th, 38th, and 39th assembly districts.
Fourteenth senate district
The combination of the 40th, 41st, and 42nd assembly districts.
Fifteenth senate district
The combination of the 43rd, 44th, and 45th assembly districts.
Sixteenth senate district
The combination of the 46th, 47th, and 48th assembly districts.
Seventeenth senate district
The combination of the 49th, 50th, and 51st assembly districts.
Eighteenth senate district
The combination of the 52nd, 53rd, and 54th assembly districts.
Nineteenth senate district
The combination of the 55th, 56th, and 57th assembly districts.
Twentieth senate district
The combination of the 58th, 59th, and 60th assembly districts.
Twenty-First senate district
The combination of the 61st, 62nd, and 63rd assembly districts.
Twenty-Second senate district
The combination of the 64th, 65th, and 66th assembly districts.
Twenty-Third senate district
The combination of the 67th, 68th, and 69th assembly districts.
Twenty-Fourth senate district
The combination of the 70th, 71st, and 72nd assembly districts.
Twenty-Fifth senate district
The combination of the 73rd, 74th, and 75th assembly districts.
Twenty-Sixth senate district
The combination of the 76th, 77th, and 78th assembly districts.
Twenty-Seventh senate district
The combination of the 79th, 80th, and 81st assembly districts.
Twenty-Eighth senate district
The combination of the 82nd, 83rd, and 84th assembly districts.
Twenty-Ninth senate district
The combination of the 85th, 86th, and 87th assembly districts.
Thirtieth senate district
The combination of the 88th, 89th, and 90th assembly districts.
Thirty-First senate district
The combination of the 91st, 92nd, and 93rd assembly districts.
Thirty-Second senate district
The combination of the 94th, 95th, and 96th assembly districts.
Thirty-Third senate district
The combination of the 97th, 98th, and 99th assembly districts.
The following territory shall constitute the 1st assembly district:
(1) Whole counties. The counties of Door and Kewaunee.
(2) Brown county. That part of the county of Brown consisting of the towns of Green Bay, Humboldt and Scott.
The following territory shall constitute the 2nd assembly district:
(1) Brown county. That part of the county of Brown consisting of:
a) the towns of Bellevue, De Pere, Eaton, Morrison and New Denmark;
b) that part of the town of Glenmore comprising ward 2;
c) the village of Denmark; and
d) that part of the city of De Pere comprising wards 3 and 6.
(2) Manitowoc county. That part of the county of Manitowoc consisting of:
a) the towns of Cato, Cooperstown, Eaton, Franklin, Gibson, Kossuth, Maple Grove, Mishicot, Rockland, Two Creeks and Two Rivers;
c) the city of Two Rivers.
The following territory shall constitute the 3rd assembly district:
(1) Brown county. That part of the county of Brown consisting of:
a) the towns of Holland, Rockland and Wrightstown;
b) that part of the town of Glenmore comprising ward 1; and
c) that part of the city of De Pere comprising wards 4 and 5.
(2) Calumet county. That part of the county of Calumet consisting of:
a) the towns of Brillion, Brothertown, Charlestown, Chilton, Harrison, Rantoul, Stockbridge and Woodville;
b) that part of the town of New Holstein comprising ward 1;
c) the villages of Hilbert, Potter, Sherwood and Stockbridge;
d) the cities of Brillion and Chilton;
e) that part of the city of Appleton located in the county; and
f) that part of the city of Menasha located in the county.
(3) Fond du Lac county. That part of the county of Fond du Lac consisting of:
a) the towns of Calumet and Marshfield;
b) that part of the town of Taycheedah comprising wards 1 and 2; and
c) the villages of Mount Calvary and St. Cloud.
(4) Outagamie county. That part of the county of Outagamie consisting of that part of the city of Appleton comprising wards 5, 8, 9 and 12.
(5) Winnebago county. That part of the county of Winnebago consisting of that part of the city of Appleton located in the county.
The following territory in the county of Brown shall constitute the 4th assembly district:
a) the village of Allouez;
b) that part of the village of Ashwaubenon comprising wards 1, 2, 5, 6, 7, 8, 9, 10, 11 and 12;
c) that part of the city of De Pere comprising wards 1, 2, 7, 8, 9, 10, 11 and 12; and
d) that part of the city of Green Bay comprising wards 41, 45, 46, 47 and 48.
The following territory shall constitute the 5th assembly district:
(1) Brown county. That part of the county of Brown consisting of:
a) the towns of Hobart and Lawrence; and
b) the village of Wrightstown.
(2) Outagamie county. That part of the county of Outagamie consisting of:
a) the towns of Buchanan, Freedom, Kaukauna, Oneida, Osborn and Vandenbroek;
b) the villages of Combined Locks, Kimberly and Little Chute; and
The following territory shall constitute the 6th assembly district:
(1) Oconto county. That part of the county of Oconto consisting of:
a) the towns of Abrams, Brazeau, Gillett, How, Maple Valley, Morgan, Oconto Falls, Spruce, Stiles and Underhill;
b) the village of Suring; and
c) the cities of Gillett and Oconto Falls.
(2) Outagamie county. That part of the county of Outagamie consisting of:
a) the towns of Bovina, Cicero, Liberty, Maine and Seymour;
b) that part of the town of Black Creek comprising ward 1;
c) the villages of Nichols and Shiocton; and
(3) Shawano county. That part of the county of Shawano consisting of:
a) the towns of Angelica, Belle Plaine, Grant, Green Valley, Hartland, Herman, Lessor, Maple Grove, Navarino, Pella, Richmond, Washington, Waukechon and Wescott;
b) the villages of Bonduel and Cecil; and
Seventh assembly district
The following territory in the county of Milwaukee shall constitute the 7th assembly district:
a) the village of West Milwaukee; and
b) that part of the city of Milwaukee comprising wards 132, 133, 137, 140, 142, 144, 148, 290, 291, 292, 295, 296, 297, 298, 299, 300, 301, 302, 303, 304, 305, 306, 307, 308, 311, 312 and 323.
The following territory in the county of Milwaukee shall constitute the 8th assembly district:
a) that part of the city of Milwaukee comprising wards 138, 139, 145, 149, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 309 and 310.
The following territory in the county of Milwaukee shall constitute the 9th assembly district:
a) that part of the city of Milwaukee comprising wards 141, 143, 146, 147, 150, 151, 152, 153, 154, 194, 197, 229, 230, 231, 232, 233, 234, 236, 237,241, 252, 253, 254, 255, 256, 257, 266, 267 and 271.
The following territory in the county of Milwaukee shall constitute the 10th assembly district:
a) that part of the village of Shorewood comprising ward 12;
b) that part of the city of Glendale comprising wards 1, 2 and 7; and
c) that part of the city of Milwaukee comprising wards 1, 19, 59, 60, 102, 103, 104, 105, 106, 107, 108, 109, 110, 113, 115, 119, 173, 174, 181, 182 and 183.
Eleventh assembly district
The following territory in the county of Milwaukee shall constitute the 11th assembly district:
a) that part of the city of Milwaukee comprising wards 2, 3, 4, 5, 8, 9, 10, 11, 12, 13, 14, 17, 18, 22, 159, 165, 166, 167, 168, 169, 170, 171, 172 and 176.
Twelfth assembly district
The following territory in the county of Milwaukee shall constitute the 12th assembly district:
a) that part of the city of Milwaukee comprising wards 20, 21, 23, 24, 25, 26, 28, 29, 33, 34, 36, 37, 38, 78, 79, 80, 157, 160, 161, 162, 163, 164, 280, 281, 283, 284 and 285.
Thirteenth assembly district
The following territory shall constitute the 13th assembly district:
(1) Milwaukee county. That part of the county of Milwaukee consisting of that part of the city of Milwaukee comprising wards 30, 31, 39, 40, 41, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 131, 286, 288 and 289.
(2) Waukesha county. That part of the county of Waukesha consisting of that part of the city of Milwaukee located in the county.
Fourteenth assembly district
The following territory in the county of Milwaukee shall constitute the 14th assembly district:
a) the city of Wauwatosa.
Fifteenth assembly district
The following territory in the county of Milwaukee shall constitute the 15th assembly district:
a) that part of the city of West Allis comprising wards 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 27, 28 and 29.
The following territory in the county of Milwaukee shall constitute the 16th assembly district:
a) that part of the city of Milwaukee comprising wards 63, 65, 66, 67, 68, 69, 111, 112, 114, 116, 117, 118, 121, 122, 134, 136, 313, 314, 315, 316, 325 and 327.
Seventeenth assembly district
The following territory in the county of Milwaukee shall constitute the 17th assembly district:
a) that part of the city of Milwaukee comprising wards 6, 7, 15, 16, 27, 32, 35, 120, 124, 125, 126, 127, 128, 130, 175, 177, 178, 179, 180, 184, 185, 186, 188, 189 and 190.
Eighteenth assembly district
The following territory in the county of Milwaukee shall constitute the 18th assembly district:
a) that part of the city of Milwaukee comprising wards 70, 71, 72, 73, 74, 75, 76, 77, 123, 129, 135, 187, 293, 294, 317, 318, 319, 320, 321, 322, 324 and 326.
Nineteenth assembly district
The following territory in the county of Milwaukee shall constitute the 19th assembly district:
a) that part of the city of Milwaukee comprising wards 43, 45, 46, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 61, 62, 64, 250, 251, 258, 259, 260, 261; 262, 263 and 265.
Twentieth assembly district
The following territory in the county of Milwaukee shall constitute the 20th assembly district:
a) the cities of Cudahy and St. Francis;
b) that part of the city of Milwaukee comprising wards 235, 238, 239, 240, 242, 243, 244, 264, 268, 269 and 270; and
c) that part of the city of South Milwaukee comprising ward 1.
Twenty-first assembly district
The following territory in the county of Milwaukee shall constitute the 21st assembly district:
a) the city of Oak Creek;
b) that part of the city of Milwaukee comprising wards 245, 246, 247, 248 and 249; and
c) that part of the city of South Milwaukee comprising wards 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 and 16.
Twenty-second assembly district
The following territory shall constitute the 22nd assembly district:
(1) Milwaukee county. That part of the county of Milwaukee consisting of:
a) the villages of Fox Point, River Hills and whitefish Bay;
b) that part of the village of Bayside located in the county;
c) that part of the village of Brown Deer comprising ward 3;
d) that part of the village of Shorewood comprising wards 1, 2, 3, 4, 5, 6, 7, 8, 9, 10 and 11; and
e) that part of the city of Milwaukee comprising wards 42, 44, 47 and 58.
(2) Ozaukee county. That part of the county of Ozaukee consisting of that part of the village of Bayside located in the county.
Twenty-third assembly district
The following territory shall constitute the 23rd assembly district:
(1) Milwaukee county. That part of the county of Milwaukee consisting of:
a) that part of the village of Brown Deer comprising wards 1, 2, 4, 5, 6, 7, 8 and 9;
b) that part of the city of Glendale comprising wards 3, 4, 5, 6, 8, 9, 10, 11 and 12; and
c) that part of the city of Milwaukee comprising wards 155, 156, 158, 272, 273, 274, 275, 277, 278, 279 and 282.
(2) Ozaukee county. That part of the county of Ozaukee consisting of that part of the city of Mequon comprising wards 11, 13, 14 and 15.
The following territory shall constitute the 24th assembly district:
(1) Washington county. That part of the county of Washington consisting of:
a) the town of Germantown;
b) that part of the town of Polk comprising wards 6 and 7;
c) that part of the town of Richfield comprising wards 1, 2, 3, 4, 6, 7, 8, 9, 10, 11, 12, 13, 14 and 15;
d) the village of Germantown; and
e) that part of the city of Milwaukee located in the county.
(2) Waukesha county. That part of the county of Waukesha consisting of:
a) that part of the village of Butler comprising wards 1 and 2; and
b) that part of the village of Menomonee Falls comprising wards 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22 and 23.
Twenty-fifth assembly district
The following territory shall constitute the 25th assembly district:
(1) Calumet county. That part of the county of Calumet consisting of that part of the city of Kiel located in the county.
(2) Manitowoc county. That part of the county of Manitowoc consisting of:
a) the towns of Centerville, Liberty, Manitowoc, Manitowoc Rapids, Meeme, Newton and Schleswig;
b) the villages of Cleveland, St. Nazianz and Valders;
c) the city of Manitowoc; and
d) that part of the city of Kiel located in the county.
Twenty-sixth assembly district
The following territory in the county of Sheboygan shall constitute the 26th assembly district:
a) the town of Sheboygan;
b) that part of the town of Sheboygan Falls comprising ward 4;
c) the village of Kohler;
d) the city of Sheboygan Falls; and
e) that part of the city of Sheboygan comprising wards 1, 2, 3, 5, 6, 9, 11, 12, 13, 14, 15 and 16.
Twenty-seventh assembly district
The following territory shall constitute the 27th assembly district:
(1) Calumet county. That part of the county of Calumet consisting of:
a) that part of the town of New Holstein comprising wards 2 and 3; and
b) the city of New Holstein.
(2) Sheboygan county. That part of the county of Sheboygan consisting of:
a) the towns of Greenbush, Herman, Lima, Lyndon, Mitchell, Mosel, Plymouth, Rhine, Russell, Scott and Wilson;
b) that part of the town of Sheboygan Falls comprising wards 1, 2 and 3;
c) the villages of Cascade, Elkhart Lake, Glenbeulah, Howards Grove and Waldo;
d) the city of Plymouth; and
e) that part of the city of Sheboygan comprising wards 4, 7, 8 and 10.
Twenty-eighth assembly district
The following territory shall constitute the 28th assembly district:
(1) Burnett county. The county of Burnett.
(2) Polk county. That part of the county of Polk consisting of:
a) the towns of Alden, Apple River, Balsam Lake, Black Brook, Bone Lake, Clam Falls, Clayton, Clear Lake, Eureka, Farmington, Garfield, Georgetown, Laketown, Lincoln, Lorain, Luck, Milltown, Osceola, St. Croix Falls, Sterling and West Sweden;
b) the villages of Balsam Lake, Centuria, Clayton, Clear Lake, Dresser, Frederic, Luck, Milltown and Osceola; and
c) the cities of Amery and St. Croix Falls.
(3) St. Croix county. That part of the county of St. Croix consisting of:
a) the town of Somerset; and
b) the village of Somerset.
The following territory shall constitute the 29th assembly district:
(1) Dunn county. That part of the county of Dunn consisting of:
a) the towns of Lucas, Menomonie, New Haven, Sheridan and Stanton;
b) the village of Knapp; and
c) the city of Menomonie.
(2) Pierce county. That part of the county of Pierce consisting of:
a) the towns of Gilman, Rock Elm and Spring Lake;
b) the village of Elmwood; and
c) that part of the village of Spring Valley located ...