Full opinion text
PER CURIAM. This case presents several challenges to the congressional and state legislative reapportionment plans enacted by the Georgia General Assembly in 2001 and 2002. Specifically, in their First Amended Complaint, the .plaintiffs contend that these plans violate a- number of constitutional and'statutory provisions, including (1) the First Amendment, (2) 2 U.S.C. § 2c, (3) Article I, § 2, based on violation of the one person, one vote principle and transgression of the state’s authority to dictate the times, places, and manner of congressional elections, and (4) the Equal Protection Clause, based on partisan gerrymandering, racial gerrymandering, violation of the one person, one vote principle, and the use of a combination of single- and multi-member districts in the state House of Representatives. Pursuant to 28 U.S.C. § 2284(a), a three-judge court, consisting of Circuit Judge • Stanley Marcus, District Judge Charles A. Pannell, Jr., and Senior District Judge William C. O’Kelley, was convened. In orders dated August 29, 2003, and October 15, 2003, we ruled on the defendant’s motions to dismiss, ultimately dismissing the plaintiffs’ 2 U.S.C. § 2 claim and the plaintiffs’ Equal Protection claim challenging the combination of single- and multi-member districts in the state House, while allowing the remaining claims to go forward. Subsequently, in an order dated December 9, 2003, we granted the defendant’s motion for summary judgment with respect to the plaintiffs’ claims of partisan gerrymandering, violation of the First Amendment, and violation of Article I, § 2 by exceeding the state’s authority to dictate the times, places, and manner of congressional elections; however, we denied both parties’ motions for summary judgment with respect to the one person, one vote claims. In that same order, we stayed consideration of the plaintiffs’ racial gerrymandering claim, which relates only to the 2001 and 2002 Senate plans, pending further development of the preclearance proceedings before the United States District Court for the District of Columbia in Georgia v. Ashcroft, Civil Action No. 01-2111. Accordingly, this three-judge district court conducted a bench trial on January 6-9, 2004, concerning only the plaintiffs’ one person, one vote challenges to Georgia’s congressional and state legislative plans. We have now considered the evidence presented at trial, as well as the parties’ deposition designations, stipulated facts, and proposed statements of fact and conclusions of law. Based upon a thorough review of the record and the applicable law, we conclude: (1) Georgia’s state legislative reapportionment plans plainly violate the one person, one vote principle embodied in the Equal Protection Clause because each deviates from population equality by a total of 9.98% of the ideal district population and there are no legitimate, consistently applied state policies which justify these population deviations. Instead, the plans arbitrarily and discrimi-natorily dilute and debase the weight of certain citizens’ votes by intentionally and systematically underpopulating districts in rural south Georgia and inner-city Atlanta, correspondingly overpopulating the districts in suburban areas surrounding Atlanta, and by underpopu-lating the districts held by incumbent Democrats. (2) Georgia’s congressional reapportionment plan, though it deviates from population equality by a maximum of seventy-two persons, does not violate Article I, § 2 because the very small population deviations are supported by legitimate state interests in avoiding additional precinct-splitting and in ensuring that those precincts that are divided are split along easily recognizable boundaries wherever possible. I. Findings of Fact A. The Reapportionment Process in General. The 2000 Decennial Census reported that the total population of the State of Georgia was 8,186,453 persons. From 1990 to 2000, the population of north Georgia, which is largely comprised of the urban and suburban areas surrounding Atlanta, grew at a much faster rate than the population of south Georgia, which is primarily rural. This population trend has remained consistent for the last several decades. In that same time period, the Republican party has also gained substantial strength in Georgia. In fact, the fastest-growing counties in the state over the past decade are Republican-leaning. Based on its substantial population growth over the previous decade, the state was entitled to two additional congressional seats pursuant to 2 U.S.C. § 2a. Because of the addition of these new congressional seats and the substantial shifts in population around the state, it became necessary for Georgia to redraw its congressional districts. See U.S. Const, art. I, § 2; 2 U.S.C. § 2a. Likewise, the state’s population growth had created population disparities in the state House and Senate districts, which needed to be corrected pursuant to Article III, § 2, ¶2 of the Georgia Constitution. The Georgia General Assembly, therefore, met in two special sessions during August and September of 2001 for the purposes of redistricting the state’s congressional seats and the state House and Senate seats. Prior to the 2001 special sessions, the House and Senate Reapportionment Committees had met both formally and informally on several occasions to prepare for the reapportionment process and to discuss various proposed plans. The committees also adopted guidelines for the reapportionment of congressional and legislative districts. These guidelines stated that the population of each state House and Senate district should be within 5% of the “ideal” district, so that the total deviation did, not exceed 10%. Plaintiffs’ Ex. 58. The guidelines also differed significantly from those adopted for the 1991 and 1981 redistrictings. The first special session of the General Assembly began on August 1 and ended on August 17, 2001. During the first special session, the General Assembly enacted a bill providing for the reapportionment of the state Senate (“the 2001 Senate Plan”). Then-Governor Roy Barnes signed the bill into law on August 24, 2001. During the first special session, the General Assembly also enacted a bill providing for the reapportionment of the state House of Representatives; however, that bill was subsequently vetoed by Governor Barnes. The second special session of the General Assembly began on August 22 and ended on September 28, 2001. During this second special session, the General Assembly enacted a second bill providing for the reapportionment of the state House of Representatives (“the House Plan”), as well as a bill providing for the reapportionment of the state’s congressional districts (“the Congressional Plan”). Governor Barnes signed both bills into law on October 1, 2001. Both houses of the General Assembly used Maptitude software to draw their redistricting plans. With the available technology and the use of this software, redistricting plans in 2001 could have been created with a deviation of 0 to 1 persons. The combination of technology and political data available to legislators and plan drafters also allowed for sophisticated analyses of political performance, so that maps could be drawn and then immediately analyzed politically. Thus, in drafting and considering their proposed maps, members of both houses relied on political performance projections, indicating the percentage of votes Democrats and Republicans would likely receive in future elections based upon an assessment of past election results. . Republicans attempted to influence the redistricting in several ways, with little success. They drew up alternative redistricting guidelines that required adherence to traditional redistricting criteria, such as district compactness and contiguity. Plaintiffs’ Ex. 59. These guidelines were rejected by both the House and Senate Reapportionment Committees. Test, of Rep. Westmoreland, Tr. at 366-68; Test, of Sen. Brown, Tr. at 662. During the special sessions, Republican legislators participated in the redistricting process by drawing up proposed plans and working with some Democratic legislators to reach compromise plans. None of these Republican-sponsored plans was passed. Test, of Sen. Johnson, Tr. at 456; Test, of Rep. Westmoreland, Tr. at 367-68, 374-75. During the special sessions, Republicans were not consulted by Democratic legislators regarding the redistricting plans. Test, of Rep. Westmoreland, Tr. at 401-02; Test, of Sen. Lee, Tr. at 511; Test, of Sen. Johnson, Tr. at 440; Test, of Bryan Tyson, Tr. at 976. Because Georgia is a jurisdiction covered by Section 5 of the Voting Rights Act, 42 U.S.C. § 1973c, it was necessary for the state to have its reapportionment plans precleared by the federal government. To that end, the state filed a complaint for declaratory judgment in the United States District Court for the District of Columbia on October 10, 2001, seeking a declaration that the plans enacted during the 2001 special sessions did not have the purpose or would not have the effect of denying or abridging the right to vote on account of race or color. In an order dated April 5, 2002, the three-judge district court presiding over that action preeleared the Congressional Plan and the House Plan, but refused to preclear the 2001 Senate Plan. See Georgia v. Ashcroft, 195 F.Supp.2d 25 (D.D.C.2002), vacated, — U.S. -, 123 S.Ct. 2498, 156 L.Ed.2d 428 (2003). At the request of the State of Georgia, the three-judge district court retained jurisdiction of the case to permit the submission of a revised Senate redistricting plan that would satisfy the requirements of Section 5. Subsequently, the General Assembly enacted a bill providing such a revised plan (“the 2002 Senate Plan”). Governor Barnes signed the bill into law on April 11, 2002, and the district court precleared the new Senate Plan on June 3, 2002. See Georgia v. Ashcroft, 204 F.Supp.2d 4, 15-16 (D.D.C.2002). The 2002 Senate Plan is not substantially different from the 2001 Senate Plan, but it does make adjustments to several districts in order to remedy the perceived violations of Section 5. Its enacting legislation specifically provides that it does “not repeal or amend the provisions of the [2001 Senate Plan]; and those provisions are merely suspended pending a final determination of their enforceability under the federal Voting Rights Act of 1965, as amended.” Meanwhile, the State of Georgia appealed the decision of the three-judge district court denying preclearance of the 2001 Senate Plan. On June 26, 2003, the United States Supreme Court vacated the district court’s decision, holding that the district court’s initial preclearance inquiry was overly narrow, and it remanded the case for further proceedings in light of its opinion. Georgia v. Ashcroft, 539 U.S. 461, 123 S.Ct. 2498, 156 L.Ed.2d 428 (2003). The district court has not yet issued a final decision on remand. Thus, because it is unclear whether the 2001 Senate Plan will ultimately be precleared and reinstated, the trial in this action considered only the 2002 Senate Plan now in effect, as well as the Congressional Plan, and the House Plan already approved by the three-judge district court in the District of Columbia. See generally Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 101, 118 S.Ct. 1003, 1016, 140 L.Ed.2d 210 (1998) (noting the Supreme Court’s long-standing disapproval of advisory opinions); Hall v. Beals, 396 U.S. 45, 48, 90 S.Ct. 200, 201-02, 24 L.Ed.2d 214 (1969) (same). B. The Individual Plans 1. The State Legislative Plans The redistricting guidelines adopted by the House and Senate Reapportionment committees indicated that “[t]he population deviation of [each] plan should not exceed an overall deviation of'10%.” Plaintiffs’ Ex. 58, at 3. Based largely on these guidelines and on the instructions given in previous redistricting cycles, 'legislators and plan drawers for both houses believed there was a “safe harbor” of ± 5% in the reapportionment of state legislative districts and, therefore, that population deviations not rising to that level did not have to bé supported by any legitimate state interest. Test, of Rep. Westmoreland, Tr. at 369-70, 409; Test, of Sen. Johnson, Tr. at 441-42; Test, of Sen. Brown, Tr. at 665-66; Test, of Linda Meggers, Tr. at 856-57. The creators of the state plans did not consider such traditional redistricting criteria as district compactness, contiguity, protecting communities of interest, and keeping counties intact. Dep. of Joe Stanton at 21, 24; Test, of Linda Meggers, Tr. at 775-76, 871-76, 718-20; Test, of Sen Brown, Tr. at 633, 637, 664-65. Rather, they had two expressly enumerated objectives: the protection of rural Georgia and inner-city Atlanta against a relative decline in their populations compared with that of the rest of the state and the protection of Democratic incumbents. a. The House Plan The House Plan consists of 180 members allocated to 147 districts, with 124 one-member districts, fifteen two-member districts, six three-member districts, and two four-member districts. In the redistricting plan immediately preceding the current plan, the House was comprised of 180 single-member districts. Based on the state’s total population according to the 2000 Census, the ideal size of a single-member House district for one person, one vote purposes is 45,480 persons. During the 2001 special sessions, the House Legislative and Congressional Reapportionment Committee, charged with coming up with a new electoral map for the House, had twenty-nine members: eighteen Democrats, including its chairman, vice-chairman, and secretary, and eleven Republicans. The chairman appointed a subcommittee, consisting entirely of Democrats, to draw the House Plan. Test, of Rep. Westmoreland, Tr. at 367. In response, the Republicans created their own subcommittee, the House Republican Subcommittee on Congressional and Legislative Reapportionment and Redistricting, and adopted the substantive redistricting guidelines that the Republicans had unsuccessfully proposed for the House to use. Test, of Rep. Westmoreland, Tr. at 367-68. The legislative reapportionment staff, and particularly Linda Meggers, worked with individual House members to draft the House Plan. In drawing the districts, Ms. Meggers took into account the political desires of various Democratic incumbents in order to achieve the ninety-one votes required to pass a plan. This was particularly difficult in south Georgia and urban Atlanta, as the districts in those areas were vastly underpopulated at the beginning of the redistricting process and the incumbents in those districts struggled to maintain as many seats as possible. Additionally, incumbents in all areas of the state sought to limit the expansion of their districts to what was considered legally necessary, i.e., a population deviation of ± 5%. Test, of Linda Meggers, Tr. at 738-40. At no time did the drafters of the plans nurture the ambition of drawing maps as close to equal in population as was reasonably practicable. In the end, many south Georgia incumbents who had seniority over other House members used them political influence to preserve the representation of rural interests as much as possible, resulting in greater negative population deviations in these areas and, consequently, greater positive population deviations in other areas. Nonetheless, south Georgia still lost seven House districts in the reapportionment plan that ultimately passed. The resulting House Plan has a total population deviation range of 9.98% and an average deviation of 3.47%. The House districts deviate from ideal equal population by a range of + 4.99% to -4.99%, with the largest district having 176,939 persons (in a four-member district) and the smallest district having 43,209 persons. Notably, ninety of the 180 House seats (50.00%) are in districts with population deviations greater than ± 4%. Sixty seats (33.33%) are in districts with deviations greater than ± 4.5%, and twenty seats (11.11%) are in districts with deviations greater than ± 4.9%. The most underpopulated districts are primarily Democratic-leaning, and the most overpopulated districts are primarily Republican-leaning. Moreover, most of the districts with negative deviations of 4% or greater are located either in south Georgia or within inner-city Atlanta. Plainly, redistricting plans could have been easily drawn with smaller population deviations; in fact, some such plans were offered for consideration but were summarily rejected. The House Plan splits eighty counties into 266 parts. The plan paired forty-two incumbents, including thirty-seven of the seventy-four incumbent Republicans (50% of the Republican caucus), but only nine of the 105 incumbent Democrats (less than 9% of the Democratic caucus). In the 2002 general election, the first general election following enactment of the plan, the composition of the House went from 105 Democrats, 74 Republicans, and 1 Independent to 107 Democrats, 72 Republicans, and 1 Independent. b. The Senate Plan The Georgia Senate consists of fifty-six members. The Georgia Constitution requires that the state senate be composed of single-member districts. Therefore, based on the state’s total population, the ideal size of a Senate district for one person, one vote purposes according to the 2000 Census is 146,187 persons. The Senate Reapportionment Committee, chaired by Senator Tim Golden, was responsible for coming up with a new electoral map in 2001. The committee was made up of twenty-four members: twenty Democrats and four Republicans. Test, of Sen. Lee, Tr. at 494. Senator Eric Johnson, the Senate’s Republican minority leader, requested a position on the committee but was denied. Test, of Sen. Johnson, Tr. at 439. Senator Golden appointed a five-member subcommittee, entirely Democratic and chaired by Senator Robert Brown, to devise a redistricting plan. Stip. Fact 28. To create a plan, Senator Brown worked with Joseph Stanton and other staff of the legislative reapportionment office, who offered technical assistance. Senator Brown focused his redistricting efforts on five primary goals: (1) drawing districts with population deviations of no greater than ± 5%; (2) ensuring that the districts did not retrogress in violation of Section 5 of the Voting Rights Act; (3) protecting or enhancing opportunities for Democrats to be elected; (4) allowing rural southern Georgia to hold on to as many seats as possible; and (5) obtaining the twenty-nine votes required to pass a plan. The resulting 2002 Senate Plan has a total population deviation of 9.98% and an average deviation of 3.78%. The Senate districts deviate from ideal equal population by a range of + 4.99% to -4.99%, with the largest district having 153,489 persons and the smallest district having 138,894 persons. Thirty-seven of the fifty-six districts (66.07%) have population deviations greater than ± 4%. Thirty-one districts (55.36%) have deviations greater than ± 4.5%, and sixteen districts (28.57%) have deviations greater than ± 4.9%. Not surprisingly, the most underpopulated districts are primarily Democratic-leaning, and the most overpopulated districts are primarily Republican-leaning. Moreover, all of the districts with negative deviations of more than 4% are situated either in south Georgia or within inner-city Atlanta. As in the House, redistrieting plans with smaller population deviations were offered for consideration, but were summarily rejected. The 2002 Senate Plan splits eighty-one counties into 219 parts. The plan also paired twelve incumbents, including ten of the twenty-four incumbent Republicans (42% of the caucus) but only two of the thirty-two incumbent Democrats (6% of the caucus). In the 2002 general election, the first general election following enactment of the plan, the composition of the Senate went from thirty-two Democrats and twenty-four Republicans to thirty Democrats and twenty-six Republicans. Following the election, four Democrats switched allegiance to the Republican Party, giving the Republicans control of the Senate by a margin of 30-26. c. The favoring of rural and inner-city interests Both the explicit admissions of witnesses for the defendant and the circumstantial evidence of the plans themselves leave no doubt that a deliberate and systematic policy of favoring rural and inner-city interests at the expense of suburban areas north, east, and west of Atlanta led to a substantial portion' of the 9.98% population deviations in both of the plans. Much of the testimony at trial centered around the distinction between “the two Georgias”: namely, the predominantly rural areas of south Georgia and the urban and suburban areas surrounding Atlanta in north Georgia. The historic struggle between these two areas of the state, with their differing views on water usage, education, transportation, economic development, and other issues, has affected Georgia politics for the last several decades and cuts across party lines. See Plaintiffs’ Ex. 205, Report of Expert Ronald Gaddie (“Gaddie Report”) at 5-7; Test, of Plaintiffs’ Expert Ronald Keith Gaddie, Tr. at 123-25; Test, of Rep. Westmoreland, Tr. at 411-12; Test, of Sen. Brown, Tr. at 621-23; Test, of Linda Meggers, Tr. at 749-51. At the same time, the population of north Georgia has increased at a significantly faster rate than that of south Georgia. Test, of Sen. Brown, Tr. at 621; Test, of Linda Meg-gers, Tr. at 729, 848-50. As a consequence, the rural areas of south Georgia have been slowly but continuously losing their political influence, particularly in the state legislature, because of their proportional loss of seats. Test, of Sen. Brown, Tr. at 621; Test, of Linda Meggers, Tr. at 729; Test, of Bryan Tyson, Tr. at 958. In an unambiguous attempt to hold onto as much of that political power as they could, and aided by what they perceived to be a 10% safe harbor, the plans’ drafters intentionally drew the state legislative plans in such a way as to minimize the loss of districts in the southern part of the state. See, e.g., Test, of Sen. Brown, Tr. at 619 (“[W]hen I looked at the southern part of the state, there was one paramount concern, and that was that we not lose any more districts than would absolutely be necessary.”); id. at 652 (“[Y]ou had the desire on the part of the rural senators to have as many rural senators as possible.”); Test, of Linda Meggers, Tr. at 739 (“[T]hey didn’t want to lose anymore representation out of rural south Georgia than they had to.”); id. at 743 (“[The negative deviations] are in south Georgia because those folks wanted to minimize their loss of power.”). In other words, the drafters redrew the majority of south Georgia’s districts, which were generally'very underpopulated going into the reapportionment process, by taking on only as much area as was needed to get within a -5% population deviation. Test, of Sen. Brown, Tr. at 651 (“[0]nce they would come within that range as far as the population is concerned, that would be as far as they would need to go. And when you consider that if you take up even more population, that means that you are constricting the number of districts that you are going to have in the south Georgia area.”); Test, of Linda Meggers, Tr. at 730 (“I took all of south Georgia and lassoed it in as if it were one big district, and we had the population, and the deviation, and how many seats. So I knew how many seats I could draw and be within five percent.”). By doing this, the plans’ drafters managed to contain the loss of seats in southern Georgia to seven seats in the House and two seats in the Senate. Test, of Linda Meggers, Tr. at 729 (“[W]ith the numbers we had, we knew that [at] a minimum they were going to lose seven seats, and my job was to keep from doing that.... They wanted me to help them see if they could draw a plan that held it to seven, if at all possible.”); id. at 752-54 (noting that seven House seats were lost in the southern part of the state); Test, of Sen. Brown, Tr. at 619 (“When I initially looked at it, I thought we were going to probably lose somewhere in the range of maybe three to four districts, and it turned out we did not lose that many.”); id. at 658 (noting that only two Senate seats were lost in the southern part of the state). A study of the patterns of deviation further illustrates how the population deviations were created. What is most striking is a comparison of the districts that were the most underpopulated and overpopulated coming into the reapportionment process and those that were the most underpopulated and overpopulated in the plans that ultimately passed. The same pattern exists in all of the charts: the most underpopulated areas are located almost exclusively in rural south Georgia and inner-city Atlanta, and the most overpopulated areas are located almost exclusively in the areas of north Georgia that encircle Atlanta. See Plaintiffs’ Exs. 3 (pre-redis-tricting Senate figures), 6 (2002 Senate Plan figures), 77 (pre-redistricting House figures), 83 (House Plan figures). Thus, it is clear that rather than using the reapportionment process to equalize districts throughout the state, legislators and plan drafters sought to shift only as much population to the state’s underpopulated districts as they thought necessary to stay within a total population deviation of 10%. And if the population trend that has transformed the state over the last several years continues, the vote dilution suffered by individuals living in significantly overpopulated districts is likely to compound over the course of this decade. d. Incumbent Protection An examination of the entire record also leads us to find that the other major cause of the deviations in both plans was an intentional effort to allow incumbent Democrats to maintain or increase their delegation, primarily by systematically underpop-ulating the districts held by incumbent Democrats, by overpopulating those of Republicans, and by deliberately pairing numerous Republican incumbents against one another. Substantial circumstantial evidence leads us to this conclusion. See, e.g., Test, of Plaintiffs’ Expert Ronald Keith Gaddie, Tr. at 46, 48-49, 81 (concluding, based on a study of the maps, that the population deviations in the state legislative plans were used for political purposes); Test, of Plaintiffs’ Expert Clark Bensen, Tr. at 239-41 (same). While Democratic incumbents who supported the plans were generally protected, Republican incumbents were regularly pitted against one another in an obviously purposeful attempt to unseat as many of them as possible. In the House Plan, forty-seven incumbents were paired, including thirty-seven Republicans, which was 50% of the Republican caucus, but only nine Democrats, comprising less than 9% of that caucus (as well as one Independent). Gaddie Report at 26, Table 8.2; Bensen Report at House App. Tab 12. Because six of the twenty-one districts involved were multi-member districts, the end result was that a maximum of twenty-eight of the paired incumbents could be reelected, and the remaining nineteen would be unseated. Gaddie Report at 26. Similarly, the 2002 Senate Plan included six incumbent pairings: four Republican-Republican pairings and two Republican-Democrat pairings. Id. at Table 8.1; Ben-sen Report at 2002 Senate Plan Section p. 6, 2002 Senate Plan App. Tab 12. In the 2002 general election, eighteen Republican incumbents in the House and four Republican incumbents in the Senate lost their seats due to the pairings, while only three Democratic incumbents in the House and no Democratic incumbents in the Senate lost seats this way. Gaddie Report at 26-27. The numbers largely speak for themselves, but the shapes of many of these districts and the resulting pairings further indicate that there was an intent not only to aid Democratic incumbents in'- getting-re-elected but also to oust many of their Republican incumbent counterparts. For example, one Republican senator (Senator Cable) was drawn into a district with a Democratic incumbent who ultimately won the 2002 general election, while an open district was drawn within two blocks of her residence. Test, of Sen. Johnson, Tr. at 444.- Additionally, two of the most senior Republican senators, Senators Burton and Ladd, were drawn into the same district, id. at 444-45; and a Republican House member, Representative Kaye, who was generally disliked by several of the Democratic incumbents, was paired with another representative in an attempt to unseat him, Test, of Linda Meggers, Tr. at 900-01. Finally, many of the districts that paired Republicans are not only oddly shaped but also overpopulated (for example, House Districts 3, 30, 46, 85, 127, and 137, which have positive deviations of 4.74%, 4.57%, 4.24%, 4.30%, 4.80%, and 4.45%, respectively, and Senate Districts 17, 40, 49, and 56, which have positive deviations of 4.97%, 4.84%, 4.61%, and 4.97%, respectively), thus also suggesting that the districts were drawn to force Republican incumbents to run against each other and to draw in as many Republican voters as possible in the process. Most of the political design in the map drawing occurred at a more basic level, with individual Democratic incumbents negotiating with the plans’ drafters to draw them the safest possible districts, i.e., districts that retained many of their previous supporters and that took on as few new constituents as was perceived to be legally necessary. Test, of Sen. Brown, Tr. at 622-26, 642, 647-50; Test, of Linda Meg-gers, Tr. at 711-12, 732, 735-40, 759-65. In fact, the evidence indicates not only that Democratic incumbents attempted to draw districts that would enhance their own prospects at re-election and further their other political ends (such as building up a support base for a future run for Congress, see, e.g., Test, of Linda Meg-gers, Tr. at 757, 759, 761, 764-66), but also that they targeted particular Republicans to prevent their re-election. Perhaps the most striking example of the manipulation of population deviations at the district level, however, may be what occurred in House District 137, which drew in two Republican incumbents. This district has a positive deviation of 4.45%, although it is located in the southern part of the state, where the vast majority of districts are underpopulated. It borders Districts 132, 133, 135, 136, 138, 139, 140, 141, and 144, which have population deviations of -4.78%, -4.63%, -4.60%, -4.68%, + 3.10%, -3.11%, -4.77%, -4.94%, and +2.21%, respectively. These efforts at selective incumbent protection through the use of population deviations and creative district shapes led to a significant overall partisan advantage for Democrats in the electoral maps. Republican-leaning districts are vastly more overpopulated as a whole than Democratic-leaning districts. Indeed, by one measure, the House Plan contains fifty overpopulated and thirteen underpopulated Republican-leaning districts, compared to only twenty-two overpopulated and fifty-nine underpopulated Democratic-leaning districts, and the 2002 Senate Plan contains nineteen overpopulated and seven underpopulated Republican-leaning districts, compared to only eight overpopulated and twenty-two underpopulated Democratic-leaning districts. Bensen Report at House App. Tab 8, 2002 Senate App. Tab 8. Moreover, as we have noted, the large positive deviations often occurred in districts that paired Republican incumbents (such as House Districts 3, 30, 46, 85, 127, and 137 and Senate Districts 17, 40, 49, and 56). e. Traditional Redistricting Criteria The Supreme Court has specifically detailed a number of state policies that, when applied in a consistent and nondiscriminatory manner, can justify some level of population deviation. In Karcher v. Daggett, 462 U.S. 725, 103 S.Ct. 2653, 77 L.Ed.2d 133 (1983), discussing population deviations, the Court indicated the kind of policies that might permit some deviation from perfect population equality: “Any number of consistently applied legislative policies might justify some variance, including, for instance, making districts compact, respecting municipal boundaries, preserving the cores of prior districts, and avoiding contests between incumbent Representatives.” Id. at 740, 103 S.Ct. at 2663. The last of these, incumbent protection, has already been discussed at length. The other policies were not causes of the population deviations in the House Plan and 2002 Senate Plan; nor indeed, were they priorities at all in drafting the plans. In fact, the defendant has never claimed that they were. First, the population deviations in the House Plan and the 2002 Senate Plan did not result from any attempt to create compact districts. One can discern that just by looking at the maps themselves — in particular, at districts such as House Districts 13, 47, 87, 127, and 137 and Senate Districts 16,17, 24, 28, and 51. Additionally, a simple comparison of these districts to the population data reveals that many of the most oddly-shaped districts in the plans have the largest positive deviations. For example, considering only the ten districts just mentioned, House Districts 13, 47, 87, 127, and 137 have positive deviations of 3.99%, 2.22%, 4.93%, 4.80%, and 4.45%, respectively; and Senate Districts 16, 17, 24, 28, and 51 have positive deviations of 4.97%, 4.97%, 4.25%, 4.97%, and 4.99%, respectively. A more sophisticated analysis of district compactness, calculated by comparing the relative length of the perimeter of a district to its area (“the Perimeter-to-Area measure”) or by measuring the space occupied by a district as a proportion of the space of the smallest encompassing circle (“the Smallest Circle score”), also demonstrates that the level of compactness is significantly smaller in the House Plan and the 2002 Senate Plan than in 1996 plans studied by plaintiffs’ expert Dr. Gaddie. Gaddie Report at 13-14. For either method of calculating compactness, a value of one indicates perfect compactness and is achieved where a district is a circle. Id. at 14. In this instance, the average Perimeter-to-Area measure for districts in the House Plan is .24, as compared to .28 in the 1996 plan; and the average Smallest Circle score for districts in the House Plan is .38, as compared to .40 in the 1996 plan. Id. at Table 4.1. An even starker contrast exists in the 2002 Senate Plan, which has an average Perimeter-to-Area measure of .16, down from .27 in the 1996 plan, and an average Smallest Circle score of .35, down from .42 in the 1996 plan. Id. The defendant does not argue that compactness was a consideration in the reapportionment process. Senator Brown, a principal architect of the 2001 and 2002 Senate Plans, never mentioned compactness as a factor in drawing districts; and Linda Meggers, the principal drafter of the House Plan, testified that redistricting decisions in Georgia are not guided by measures of compactness. Test, of Linda Meggers, Tr. at 775-76, 875-76. Other witnesses similarly testified thát compactness was not considered in drawing the maps. See. e.g., Dep. of John G. Kirincich, Jr. at 76, 216; Dep. of Douglas M. Moore at 30; Dep. of Joe Stanton at 24, 50-51, 93, 99, 106-07. Nor was compactness specifically mentioned in the guidelines for redistricting adopted by the House and Senate Reapportionment Committees. Plaintiffs’ Ex. 58. Second, district contiguity was not a real concern among plan drafters and legislators. See, e.g., Test. of Sen. Brown, Tr. at 664-65 (stating that he was “never quite sure” what contiguity meant); Test, of Linda Meggers, Tr. at 871-73 (indicating that the definition of “contiguous” was relaxed prior to the 2001 reapportionment process, thus allowing for the use of point contiguity). It is also clear that any regard plan drafters may have had for contiguity could not explain the 9.98% population deviations in the plans. While all of the districts are technically contiguous (as required by state law), many districts achieve that designation through the use of water contiguity, which is predicated on the assumption of line-of-sight across a lake or other body of water, or touch-point contiguity, which is predicated on facing corners in a checker-board like fashion. Water contiguity was necessary to keep together six districts in the House Plan (specifically, Districts 77, 127, 128, 129, 141, and 146) and seventeen districts in the 2002 Senate Plan (specifically, Districts 1, 3, 11, 12, 13, 14, 18, 19, 20, 21, 25, 26, 29, 45, 49, 51, and 54). Touch-point contiguity is found in five districts in the House Plan (specifically, Districts 3, 72, 96, 114, and 120) and in one district in the 2002 Senate Plan (specifically, District 18). Gaddie Report at 15-16, Tables 5.1, 5.2; Report of Plaintiffs’ Expert Clark Bensen, Plaintiffs’ Ex. 204 (“Bensen Report”), at 3-5. Furthermore, the majority of the districts that are contiguous only by reason of water or touch-point contiguity are overpopulated: sixteen of the twenty-three districts with water contiguity have deviations of more than ± 4%, and two of the six districts with touch-point contiguity have deviations of more than ± 4.5%. Gaddie Report at 16. Third, it is also clear that the drafters of the plans were almost entirely unconcerned about keeping counties whole, and that the 9.98% total population deviations cannot be explained by efforts to keep counties together. Both Senator Brown and Linda Meggers testified that as the state has become more diverse and more mobile, county lines have become less and less relevant on a statewide level. Test, of Sen. Brown, Tr. at 633 (“Georgia is a much more mobile state. The old county unit system that once existed has less relevance to state policy, state governmental interests than it once did.”); id. at 637 (stating that “I do not share [the] opinion” that splitting counties negatively impacts local interests); Test, of Linda Meggers, Tr. at 718-20 (stating that counties are becoming less important as political and social units than they used to be). A study of the maps also reveals that their drafters were not concerned about splitting counties. The House Plan actually splits eighty of the state’s 159 counties into 266 parts. In contrast, the 1996 plan split only seventy-two counties, even though that plan contained 180 districts and the current plan contains only 147 (due to the use of multi-member districts). Moreover, in the 2002 Senate Plan, eighty-one counties are split into 219 parts, whereas the 1996 plan split only forty-three counties. In fact, some counties are split into multiple parts, such as Cobb County (thirteen parts in the House Plan and six parts in the 2002 Senate Plan), DeKalb County (twelve parts in the House Plan and seven parts in the 2002 Senate Plan), Fulton County (sixteen parts in the House Plan and eight parts in the 2002 Senate Plan), and Gwinnett County (nine parts in the House Plan and seven parts in the 2002 Senate Plan). Though part of this division is a result of highly populated counties that cannot be entirely encompassed within a single district, some counties are split into multiple parts even in the more rural areas of the state. For example, Bartow County, which has 75,000 residents, is split into three House Districts and four Senate Districts; and Bryan County, which has just 23,417 residents, is split into five House Districts and three Senate Districts. Gaddie Report at 19, 20, Tables 6.2, 6.3. Additionally, forty-two of the counties split in the House Plan, and sixty-eight of counties split in the 2002 Senate Plan, have populations smaller than the ideal district size. Id. at 19. There is no correlation between county splits and attempts to reduce population disparities, as sixty-five of the counties split in the House Plan and seventy-one of the counties split in the 2002 Senate Plan contain at least one district with a population deviation of ± 4% or greater. Id. Moreover, many of these high-deviation districts are located adjacent to other districts whose populations could be transferred so as to reduce the deviations in both districts and, in some cases, to eliminate the county split at the same time. Id; Test, of Plaintiffs’ Expert Ronald Keith Gaddie, Tr. at 70-71. Finally, preserving the cores of prior districts was not mentioned in the guidelines for redistricting adopted by the House and Senate Reapportionment Committees. Plaintiffs’ Ex. 58. To the extent that the preservation of district cores may have been a goal of the drafters of the House and Senate Plans, they tried only to maintain the cores of Democratic-leaning districts. The plan drafters drew districts for Democratic incumbents who supported the plans, allowing them to retain as much of their base as they wanted and to take on as little new territory as possible, and simply gave Republican incumbents what was left, i.e.', “remnant districts.” Test, of Sen. Brown, Tr. at 623-26, 634, 647-49; Test, of Linda Meggers, Tr. at 712, 740, 760, 905-06. The plans’ core retention figures, which represent the extent to which constituencies are maintained or disrupted by a new map, amply demonstrate that core retention was not a concern in the redistricting process. Core retention can be viewed in one of two ways: (1). in terms of the largest core of a prior district that is included in a successor district, or (2) in terms of the district core of each incumbent located in a district. Gaddie Report at 22-23. In the House Plan, the average largest district core is 70.84% for single-member districts, 42.97% for two-member districts, 28.27% for three-member districts, and 23.80% for four-member districts; and the average incumbent core is 65.88% for single-member districts, 32.18% for two-member districts, 22.32% for three-member districts, and 23.72% for four-member districts. Id. at 23-24, Table 7.3. In the 2002 Senate Plan, the average largest core is 63.39%, and the average incumbent core is 58.22%. Id. at 23, Table 7.2. To the extent that the map drawers had any concern for retaining the cores of pri- or districts, a comparison of Republican versus Democrat core retention reveals the extremely inconsistent manner in which this factor was applied. Despite the fact that Republican areas of the state have grown at a faster rate than have Democratic areas (so that in a neutral plan Republican incumbents would be expected to take on fewer new constituents), core retention figures were significantly lower — especially in terms of incumbent cores — for.Republican than for Democratic incumbents. In the House, Republicans in single-, two-, and three-member districts retained an average of 56.73%, 28.36%, and 17.95% of their core districts, respectively, whereas Democrats retained 73.93%, 38.48%, and 27.56%, respectively. Id. at 24, Table 7.3. Similarly, in the Senate, Republicans retained an average of 45.67% of their core districts, while Democrats retained 66.60%. Id. at 23, Table 7.2. . After thorough review of the entire record in this case, we cannot escape the conclusion that the population deviations were designed to allow Democrats to maintain or increase their representation in the House and Senate through the underpopulation of districts in Democratic-leaning rural and inner-city areas of the state and through the protection of Democratic incumbents and the impairment of the Republican incumbents’ reelection prospects. The twin goals of regional- favoritism and protection of Democratic incumbents led to the underpopulation and overpopulation of certain districts. Rural and inner-city Atlanta districts tended also to be Democratic-leaning and to be represented by Democrats, so the negative population deviations in these districts were ultimately the result of design. Likewise, suburban districts in northern Georgia tended to be Republican-leaning and to have Republican representation in the General Assembly, and so these districts were overpopulated. 2. The Congressional Plan Prior to the 2000 census, the State of Georgia had eleven congressional districts. Those districts were drawn in 1995 by a three-judge court in the Southern District of Georgia after the Georgia General Assembly was unable to fashion a constitutional plan. Johnson v. Miller, 922 F.Supp. 1556 (S.D.Ga.1995) (three-judge court), aff'd sub nom. Abrams v. Johnson, 521 U.S. 74, 117 S.Ct. 1925, 138 L.Ed.2d 285 (1997). The range in population deviation for the court-drawn plan was from -0.23% in District 2 to + 0.12% in District 3, for an overall deviation of 0.35% and an average deviation of 0.11%. The 2000 census figures indicated that the state had gained more than 1.7 million people in the preceding ten years, resulting in the allocation of two additional congressional seats. With a population of 8,186,453 people, Georgia was entitled to thirteen congressional districts, each with an ideal population of 629,727 persons. Congressional redistricting in Georgia is a collaborative effort between the state House of Representatives and the state Senate. The House and Senate each passed a redistricting plan, both of which were referred to a conference committee composed of six members, three each from the House and Senate. There were no Republican representatives on the conference committee. On the final day of the special session, the conference committee sent a compromise redistricting plan back to the House and Senate. Each chamber passed the plan, and it was signed by the governor. The new Congressional Plan resulted in a total population deviation of seventy-two people, approximately 0.01% of the population of an ideal congressional district. The most overpopulated district, District 9, has thirty-five more people than the perfectly apportioned district, and the most underpopulated district, District 4, has thirty-seven fewer people than the perfectly apportioned district. The only district that is within one person of the ideal number is District 5, with a population deviation of zero. The average district’s absolute deviation was seventeen people, or 0.003% of an ideal district. Senator Brown testified that deviations, and their correlation with political advantage, were not considered when the congressional maps were created. Indeed, it is difficult to imagine how deviations this small could amount to a significant advantage for any party. After the 2002 elections, the first held under the new plan, Republicans controlled eight of the congressional seats compared to the Democrats’ five seats. The deviations that exist in the Congressional Plan could have been reduced or eliminated all together. Linda Meggers, who was directly involved with drafting both the House plan that was sent to the conference committee and the final plan that emerged from the conference committee, testified that it would be possible to draw a congressional map for the State of Georgia with a population deviation of plus or minus one person. Test, of Linda Meg-gers, Tr. at 881. In fact, such a map could have been created in one day, even while ensuring compliance with the Voting Rights Act. Id. at 881-82. Ms. Meggers further testified that she could create such a plan that split fewer counties than the present plan, is more compact than the present plan, and divides fewer voting precincts than the present plan. Id. at 882-83. It is clear that politics and the individual concerns of various legislators had an impact on the congressional redistricting process. Ms. Meggers testified that former Senate Majority Leader Walker insisted on a district in east Georgia that he hoped would lead to a successful congressional campaign for his son. Id. at 781-83. The resulting district, District 12, runs from Clarke County down the eastern border of the state to Bryan County. Id. at 783. Ms. Meggers, also testified about the political battles that led to the creation of District 13, an irregular-shaped district that touches parts of eleven counties surrounding the city of Atlanta. District 13 resulted from Senator Hecht’s desire to create a district from which he felt he could mount a successful congressional campaign. Id. at 782-87. As more senators realized that Senator Hecht was attempting to create such a district, they too began to assert influence on the proposed District 13, drawing in different areas they felt would be advantageous in the event they attempted to launch them own congressional careers. Id. at 784-87. These considerations drove not only the shape of District 13 but also the shape of the entire map. Id. As each vote in the Senate would be of critical importance, the drafters could not afford to alienate any one senator by disregarding his or her personal desires. Id. Ms. Meggers further testified that representatives from both the House and the Senate wanted a “middle Georgia” congressional district, designed to protect the interests of groups from that area of the state. Id. at 782-88. That district ended up being District 3, which runs across the middle of the state, from Marion County in the west to Evans County in the east. Id. at 788. Former Speaker of the House Murphy, who was represented by Congressman Barr in the former District 7, wanted a new district drawn in that area that would be more competitive for Democrats and more challenging for Congressman Barr. Id. at 789-90. Senator Eric Johnson, a Republican, similarly testified that passing a congressional plan is an extraordinarily political process because so many legislators have aspirations of being elected to Congress and, therefore, have an interest in crafting a district they consider politically desirable. Test, of Sen. Johnson, Tr. at 463. Senator Johnson stated that during the redistrictmg process, nobody mentioned a desire to avoid splitting counties or precincts; there was no real effort to keep communities of interest together; there was no discussion of making compact districts; nobody discussed having district or precinct lines follow natural or man-made boundaries; and there was no effort to protect incumbents. Id. at 464-65. Senator Johnson gave testimony substantially similar to that of Ms. Meggers regarding Senator Walker’s desire to craft a congressional district from which his son might pursue a campaign and Speaker Murphy’s desire to draw himself out of Congressman Barr’s district. Id. at 468, 470. After the politicians crafted a version of the congressional map, Ms. Meggers and her staff were left to work out the details. The State of Georgia did not reduce the population deviations to zero because to do so would have required either splitting more precincts or further splitting existing split precincts along something other than an easily recognizable boundary. Test, of Linda Meggers, Tr. at 804-07. In constructing the final map, when Ms. Meggers had to split a voting precinct, she tried to do so along only an obvious natural or man-made boundary. Id. For example, she would try to split the precinct along a major road as opposed to a neighborhood street. This was done for several reasons. First, having precinct lines correspond with major natural or man-made boundaries made it easier for election officials who are responsible for maintaining an accurate list of voters. Id. at 805. Second, it is easier for voters to determine what district they are in when easily recognizable boundaries are used. Id. Finally, she testified to a concern regarding ballot secrecy. Id. at 805-06. The secrecy issue relates to the fact that precincts are split for a variety of different districts, including congressional districts, Senate districts, House districts, school board districts, county commission districts, etc. Ms. Meggers expressed a concern about creating ballot combinations belonging to such a small number of people that one could determine a voter’s identity simply by knowing the different districts within which that person voted. Id. Ms. Meggers explained in detail what she would have had to do to reduce each district to a population deviation of plus or minus one person. For example, she testified that District 2, which is located in southwest Georgia, had only three split precincts, all in the Columbus area of Mus-cogee County. It had a population deviation of plus eight people. In order to reduce the deviation to zero, she would need to find a census block in one of those three precincts that bordered neighboring District 11 and that had exactly eight people. While one such voting block existed, as depicted in Defendant’s Exhibit 6E, it would have required deviating from the major boundary, Edgewood Road, that was used to divide the precinct. However, even if she made such a change, that would have resulted in a change in the deviation in the adjoining District 11. Ms. Meggers explained that the only way to correct the deviation in District 11 was to deviate from a major boundary in an already split precinct; and even if that were done, it would cause a deviation that would need to be corrected in District 8. Ms. Meggers continued with the explanations, demonstrating how each deviation correction in each district could be accomplished only by departing from a major boundary and even then would serve only to create a deviation fluctuation in an adjoining district. II. Conclusions of Law The Constitution of the United States requires that congressional and state legislative seats be apportioned equally, so as to ensure that the constitutionally guaranteed right of suffrage is not denied by debasement or dilution of the weight of a citizen’s vote. Reynolds v. Sims, 377 U.S. 533, 555, 568, 84 S.Ct. 1362, 1378, 1385, 12 L.Ed.2d 506 (1964); Wesberry v. Sanders, 376 U.S. 1, 7-8, 84 S.Ct. 526, 530, 11 L.Ed.2d 481 (1964). As the Supreme Court expressed more than forty years ago in a challenge to one of Georgia’s previous state legislative reapportionment schemes: Once the geographical unit for which a representative is to be chosen is designated, all who participate in the election are to have an equal vote — whatever their race, whatever their sex, whatever their occupation, whatever their income, and wherever their home may be in that geographical unit. This is required by the Equal Protection Clause of the Fourteenth Amendment. The concept of “we the people” under the Constitution visualizes no preferred class of voters but equality among those who meet the basic qualifications. The idea that every voter is equal to every other voter in his State, when he casts his ballot in favor of one of several competing candidates, underlies many of our decisions. Gray v. Sanders, 372 U.S. 368, 379-80, 83 S.Ct. 801, 808, 9 L.Ed.2d 821 (1963). While the Court has allowed some flexibility in state legislative reapportionment and, to a lesser extent, in congressional reapportionment, the central and invariable objective in both instances remains “equal representation for equal numbers of people.” Wesberry, 376 U.S. at 18, 84 S.Ct. at 535; see also Reynolds, 377 U.S. at 579, 84 S.Ct. at 1390. Thus, deviations from exact population equality may be allowed in some instances in order to further legitimate state interests such as making districts compact and contiguous, respecting political subdivisions, maintaining the cores of prior districts, and avoiding incumbent pairings. See Karcher v. Daggett, 462 U.S. 725, 740-41, 103 S.Ct. 2653, 2663, 77 L.Ed.2d 133 (1983) (holding that State failed to meet its burden of proving deviations were necessary to achieve legitimate, nondiscriminatory legislative policy); Reynolds, 377 U.S. at 578-79, 84 S.Ct. at 1390-91. However, where population deviations are not supported by such legitimate interests but, rather, are tainted by arbitrariness or discrimination, they cannot withstand constitutional scrutiny. See Roman v. Sincock, 377 U.S. 695, 710, 84 S.Ct. 1449, 1458, 12 L.Ed.2d 620 (1964). The population deviations in the Georgia House and Senate Plans are not the result of an effort to further any legitimate, consistently applied state policy. Rather, we have found that the deviations were systematically and intentionally created (1) to allow rural southern Georgia and inner-city Atlanta to maintain their legislative influence even as them rate of population growth lags behind that of the rest of the state; and (2) to protect Democratic incumbents. Neither of these explanations withstands Equal Protection scrutiny. First, forty years of Supreme Court jurisprudence have established that the creation of deviations for the purpose of allowing the people of certain geographic regions of a state to hold legislative power to a degree disproportionate to their population is plainly unconstitutional. Moreover, the protection of incumbents is a permissible cause of population deviations only when it is limited to the avoidance of contests between incumbents and is applied in a consistent and nondiscriminatory manner. The incumbency protection in the Georgia state legislative plans meets neither criterion. Therefore, that interest cannot save the plans from constitutional infirmity. Quite simply, the Georgia plans violate the Equal Protection Clause. “Full and effective participation by all citizens in state government requires ... that each citizen have an equally effective voice in the election of members of his state legislature. Modern and viable state government needs, and the Constitution demands, no less.” Reynolds, 377 U.S. at 565, 84 S.Ct. at 1383. By contrast, we are satisfied that Georgia’s congressional redistricting plan passes constitutional muster, as its minimal seventy-two person deviation appears to us to have been caused by legitimate efforts to limit the number of precinct splits and to contain any necessary precinct splits to easily recognizable boundaries. We are in no way unmindful that in striking down two redistricting plans, we are necessarily interfering with the legislative process of reapportionment. The Supreme Court has recognized that the goal of fair and effective representation is not furthered “by making the standards of reapportionment so difficult to satisfy that the reapportionment task is recurringly removed from legislative hands and performed by federal courts.” Gaffney v. Cummings, 412 U.S. 735, 749, 93 S.Ct. 2321, 2329, 37 L.Ed.2d 298 (1973); see also Miller v. Johnson, 515 U.S. 900, 915, 115 S.Ct. 2475, 2488, 132 L.Ed.2d 762 (1995) (“Federal-court review of districting legislation represents a serious intrusion on the most vital of local functions. It is well settled that ‘reapportionment is primarily the duty and responsibility of the State.’” (quoting Chapman v. Meier, 420 U.S. 1, 27, 95 S.Ct. 751, 766, 42 L.Ed.2d 766 (1975))). However, where a state’s reapportionment intrudes upon the fundamental right to vote for what can be characterized only as discriminatory and arbitrary reasons, it is our duty to step in. As the Reynolds Court advised: We are told that the matter of apportioning representation in a state legislature is a complex and many-faceted one. We are advised that States can rationally consider factors other than population in apportioning legislative representation. We are admonished not to restrict the power of the States to impose differing views as to political philosophy on their citizens. We are cautioned about the dangers of entering into political thickets and mathematical quagmires. Our answer is this: a denial of constitutionally protected rights demands judicial protection; our oath and our office require no less of us.... “When a State exercises power wholly within the domain of state interest, it is insulated from federal judicial review. But such insulation is not carried over when state power is used as an instrument for circumventing a federally protected right.” 377 U.S. at 566, 84 S.Ct. at 1384 (quot