*1308 TABLE OF CONTENTS
I. Introduction.............................................................1308
II. Procedural History.......................................................1310
III. Evidentiary Background..................................................1311
A. Texas Demography Related to Redistricting............................1311
B. Pertinent History Related to Redistricting in Texas.....................1312
C. The 1991 Congressional Redistricting Process..........................1313
1. General Background..............................................1313
2. Voting Rights Act Considerations..................................1314
a. Racial Polarization............................................1316
b. History of Discrimination......................................1317
3. Incumbents’ Interests.............................................1317
4. Use of Racial Data...............................................1318
5. Congressional District 30..........................................1319
6. Congressional Districts 18 and 29.................................. 1323
7. Congressional District 28.......................................... 1325
8. Other Congressional Districts......................................1326
D. Expert Testimony....................................................1328
E. Other Districting Plans...............................................1330
TV. Factual Findings and Legal Conclusions...................................1331
A. The Voting Rights Districts...........................................1337
1. Congressional District 30.......................................... 1337
2. Congressional Districts 18 and 29.................................. 1339
3. Narrow Tailoring to Achieve a Compelling State Interest?...........1341
4. Congressional District 28.......................................... 1344
B. Other Congressional Districts.........................................1344
V. Conclusion...............................................................1345
Special Concurrence......................................................1345
Appendix (Maps of Districts 18, 29, 30).................................... 1348
Order...................................................................1351
Before JONES, Circuit Judge, HITTNER and HARMON, District Judges.OPINION
EDITH H. JONES, Circuit Judge:I. INTRODUCTION
The Voting Rights Act of 1965 at one blow demolished the obvious devices that southern states had used to disenfranchise African-American voters for decades. The Act marked the full maturity in American political life of the Founders’ idea that “all men are created equal” and the Rev. Martin Luther King’s hope that his children would be judged by the content of their character, not the color of their skin. The meaning of equality—as also enshrined in the Fourteenth Amendment’s guarantee of “equal protection of the laws”—is the subject of this lawsuit.
It is no longer disputed that the Fourteenth and Fifteenth Amendments embody a right to ballot box equality among American citizens of different races or ethnic backgrounds. See, e.g., Rogers v. Lodge, 458 U.S. 613, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982); Baker v. Carr, 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962); Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960). The Fourteenth Amendment also prohibits government from invidiously classifying persons because of their race. Repeatedly and in the strongest terms, the Supreme Court has condemned intentional racial discrimination by state agents or bodies. Where official discrimination is found to exist, the burden is on the governmental body to justify it by no less than a compelling governmental interest.
One year ago, the Supreme Court reaffirmed that intentional racial discrimination is offensive to the Equal Protection Clause when it occurs as part of legislative redistricting. See Shaw v. Reno, — U.S. -, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). In Shaw, the Court held that “redistricting legislation [is unconstitutional if it] is so extremely irregular on its face that it rationally can be viewed only as an effort to segregate *1309the races for purposes of voting, without regard for traditional districting principles and without sufficiently compelling justification.” Id. at-, 113 S.Ct. at 2824.
In 1991, the State of Texas deliberately redrew its Congressional boundary lines following the 1990 census with nearly exact knowledge of the racial makeup of every inhabited block of land in the state. This insight, worthy of Orwell’s Big Brother, was attainable because computer technology, made available since the last decennial census, superimposed at a touch of the keyboard block-by-block racial census statistics upon the detailed local maps vital to the redistricting process. Not only did the state know the precise location of African-American, Hispanic, and Anglo populations, but it repeatedly segregated those populations by race to further the prospects of incumbent officeholders or to create “majority-minority” Congressional districts. The result of the Legislature’s efforts is House Bill 1 (“HBl”), a crazy-quilt of districts that more closely resembles a Modigliani painting than the work of public-spirited representatives.1
The challenged plan (HBl) was passed in the second called session of the 72nd Texas Legislature and signed into law by the Governor on August 29,1991. See Plaintiff Exh. 1. On November 18, 1991, the Texas Congressional Redistricting Plan received § 5 preelearance from the Attorney General.2 See United States Exh. 1007; Stip. 37. Notwithstanding the preclearance, the Attorney General expressed fundamental reservations about the redistricting plan:
While we are preclearing this plan under Section 5, the extraordinarily convoluted nature of some districts compels me to
disclaim any implication that the proposed plan is otherwise lawful or constitutional.
United States Exh. 1007 at 2.
The plaintiffs in this case are six Texas voters who reside in Congressional Districts 18, 25, 29, and 30. In a pretrial stipulation, they alleged that 24 of the state’s 30 Congressional Districts are the product of racial gerrymandering or intentional racial discrimination.3
The question before this court is whether any of the 24 challenged Congressional Districts, many of whose boundaries were clearly affected by racial considerations, can be sufficiently explained by legitimate redistricting criteria other than race. See Shaw, — U.S. at -, 113 S.Ct. at 2824. For reasons that follow, we conclude that Congressional Districts 18, 29, and 30 as presently drawn are not so explainable. They were conceived for the purpose of providing “safe” seats in Congress for two African-American and an Hispanic representatives. They were scientifically designed to muster a minimum percentage of the favored minority or ethnic group; minority numbers are virtually all that mattered in the shape of those districts. Those districts consequently bear the odious imprint of racial apartheid, and districts that intermesh with them are necessarily racially tainted.
Other challenged Texas Congressional Districts are disfigured4 less to favor or disadvantage one race or ethnic group than to promote the reelection of incumbents; they are not unconstitutionally segregated.
We do not hold that the state may only draw Congressional boundaries with a blind eye toward race, a goal which would be impossible, nor that it is altogether prohibited from creating majority-minority districts. *1310But when the State redraws the boundaries of Districts 18, 29, and 30 and contiguous districts, it can and must exhibit respect for neighborhoods, communities, and political subdivision lines. As the Supreme Court put it, appearances do matter. Id. at-, 113 S.Ct. at 2827. In appearance and in reality, these three districts were racially gerrymandered.
Racial gerrymandering is unconstitutional, but it is also morally wrong, inconsistent with our founding tradition and Martin Luther King’s vision. The color of a person’s skin or his or her ethnic identity is the least meaningful way in which to understand that person. To elevate racial classification as a basis for political representation inevitably defeats the principle of equality because it causes all of society to become more, not less, race-conscious. Justice William 0. Douglas put this point well:
When racial or religious lines are drawn by the State, the multiracial, multireligious communities that our Constitution seeks to weld together as one become separatist; antagonisms that relate to race or to religion rather than to political issues are generated; communities seek not the best representative but the best racial or religious partisan. Since that system is at war with the democratic ideal, it should find no footing here.
Wright v. Rockefeller, 376 U.S. 52, 67, 84 S.Ct. 603, 611, 11 L.Ed.2d 512 (1964) (Douglas, J., dissenting) (quoted in Shaw, — U.S. at -, 113 S.Ct. at 2827).
II. PROCEDURAL HISTORY
The plaintiffs are six registered voters who reside in Congressional Districts 18, 25, and 29 (located in whole or in part in Harris County), and in District 30 (most of which is located in Dallas County). See Complaint at 4 ¶7. Plaintiffs filed their Original Complaint for Permanent Injunction and Declaratory Judgment and Motion for Preliminary Injunction on January 26, 1994 against the Governor, the Lieutenant Governor, the Attorney General, and the Secretary of State as well as the Speaker of the Texas House of Representatives.
The complaint alleged that the 1991 Congressional Redistricting Plan for the State of Texas “represents an unconstitutional effort to segregate the races for purposes of voting: (1) without regard for traditional districting principles, including compactness, contiguousness [sic], consistency with existing political, economic, societal, governmental or jurisdictional boundaries; (2) without sufficiently compelling justification; and (3) without ‘narrow tailoring’ as required by the United States Constitution.” Complaint at 2 ¶ l.5
Candidate qualifying for the March 8, 1994 primary elections in Texas closed on January 3, 1994 and early voting began on February 16. On March 2, 1994, the court enteied an order denying the plaintiffs’ motion for a preliminary injunction and their motion for consolidation and expedited hearing and set trial for June 28, 1994. Also on March 2, 1994, the court granted the motion of the United States to participate as amicus curiae in the case. On March 14,1994, the state defendants in this action filed their answer to the complaint.
On May 5, 1994, the court granted the motion to intervene of six African-American registered voters represented by the NAACP Legal Defense and Educational Fund, Inc. (“Lawson Intervenors”). On May 20, 1994, the court granted the motion of the United States to intervene. A week later, on May 27, 1994, the court entered an order granting intervention to both The League of United Latin American Citizens (“LULAC”) and seven Hispanic registered voter members of the organization.
On June 13,1994, the United States filed a motion to bifurcate trial; the court denied the motion on June 17, 1994. On June 16, 1994, the court conducted the pretrial conference. At the pretrial conference, the court set the pretrial schedule and directed the plaintiffs to file a statement narrowing the districts to which they asserted challenges *1311and eliminating any claims not supported by substantial evidence or case law. In response, on June 16,1994, the plaintiffs filed a statement dismissing their § 26 and state constitutional claims and identifying six districts that they did not challenge under the Fourteenth Amendment. In a subsequent filing, the plaintiffs dismissed their Fifteenth Amendment claims.7 Trial was held June 27-30 and concluded on July 1, 1994.
To expedite matters, this court limited the parties’ trial time while permitting them to submit virtually unlimited additional documentary and deposition evidence. The parties liberally accepted this offer.8 The court has reviewed all of the evidence brought before us. The record references below highlight and summarize the testimony.
III. EVIDENTIARY BACKGROUND
A. Texas Demography Related to Redistricting
Congressional redistricting in Texas operated against a backdrop of important demographic changes throughout the state. Population growth from 1980 to 1990 was largely attributable to significant population growth among Hispanics and African-Americans. Of particular interest is the enormous increase in Hispanic population state-wide. Thus, what follows are the Census figures chronicling minority-led population growth in Texas during the 1980’s.
According to the 1980 Census, Texas’ total population was 14,229,191, of whom 2,985,824 (20.98%) were Hispanic, 1,692,542 (11.89%) were non-Hispanic African-American, and 9,350,297 (65.7%) were Anglo. See Stip. 7. By the 1990 Census, Texas’ total population had increased to 16,986,510, of whom 4,339,-905 (22.55%) were Hispanic, 1,976,360 (11.-63%) were non-Hispanic African-American, and 10,291,680 (60.59%) were Anglo. The increase in population from 1980 to 1990 (2,757,319 persons) entitled Texas to three additional seats in the United States House of Representatives, increasing the size of the delegation from 27 to 30. See Stip. 8. Based on the 1990 Census, the ideal size of a Texas Congressional district is 566,217. See Stip. 17.
Under the 1980 Census, Texas’ voting-age population was 9,923,085, of whom 1,756,971 (17.71%) were Hispanic, 1,095,836 (11.04%) were non-Hispanic African-American, and 6,932,894 (69.87%) were Anglo. See Stip. 9. By 1990, Texas’ voting-age population had increased to 12,150,671, of whom 2,719,586 (22.38%) were Hispanic, 1,336,688 (11.0%) were non-Hispanie African-American, and 7,828,352 (64.43%) were Anglo. See Stip. 10. Taking citizenship into account alters these percentages. Under 1990 figures, the total citizen voting age population is only 11,313,-641, of whom 2,085,857 (18.4%) were Hispanic, and 1,315,860 (11.6%) were non-Hispanie African-American. See State Exh. 14, Appendix 1.
Even a cursory review of the foregoing Census data reveals the significant growth experienced by minority communities, and, in particular, the explosive population growth among Hispanics in Texas. The Hispanic population in the state grew from 1980 to 1990 by 1,354,081 persons, or 45.4%; the African-American population in the state grew from 1980 to 1990 by 283,818 persons, or 16.8%; and the Anglo population in the *1312state grew from 1980 to 1990 by 941,383 persons, or 10.1%. The growth in Hispanic population accounted for a remarkable 49.1% of the increase in Texas’ total population from 1980 to 1990. See Stip. 11.
The four counties "with the largest growth in population from 1980 to 1990 in number of persons gained are Harris, Tarrant, Dallas, and Bexar Counties. As noted below, the growth in the Hispanic and African-American population in these counties accounted for a significant proportion of the increase in population in each of these counties:
a. The total population of Harris County increased by 408,652 persons from 1980 to 1990. The Hispanic population in the county increased by 275,858 persons, accounting for 67.5% of the growth in total population in the county. The African-American population increased by 58,674 persons, accounting for 14.4% of the county’s growth. See Stip. 12. According to the 1990 Census, there are 644,935 Hispanic persons in Harris County, of whom 405,735 are of voting age. See Stip. 13. Furthermore, there are 527,964 African-American persons in the county, of whom 359,248 are of voting age. See Stip. 14.
b. The total population of Tarrant County increased by 309,223 persons from 1980 to 1990. The Hispanic population in Tarrant County increased by 72,247 persons, accounting for 23.4% of the growth in total population in the county. The African-American population in Tarrant County increased by 37,765 persons, accounting for 12.2% of the county’s growth. See Stip. 12.
c. The total population of Dallas County increased by 296,420 persons from 1980 to 1990. The Hispanic population in the county increased by 161,069 persons, accounting for 54% of the growth in the total population of the county. The African-American population increased by 76,343 persons, accounting for 25.8% of the county’s growth. See Stip. 12. In Dallas County, there are 362,130 African-American persons of whom 243,918 are of voting age. See Stip. 15.
d.The total population of Bexar County increased by 196,594 persons from 1980 to 1990. The Hispanic population in Bexar County increased by 128,269 persons, accounting for 65.2% of the growth in the total population of the county. The African-American population in Bexar County increased by 13,326 persons, accounting for 6.8% of the county’s growth. See Stip. 12.
Significant increases in Hispanic population occurred between 1980 and 1990 in several other counties:
a. The Hispanic population of Cameron County increased by 51,341 persons between 1980 and 1990.
b. The Hispanic population of Hidalgo County increased by 96,760 persons between 1980 and 1990.
c. The Hispanic population of Webb County increased by 34,227 persons between 1980 and 1990.
See Stip. 16.
B. Pertinent History Related to Redistricting in Texas
Texas did not redistrict Congressional Districts at all between 1933 and 1957. See United States Exh. 1071 at 7. Following the 1960 Census, the state “redistricted” by creating a new at-large Congressional seat. See State Exh. 23, ¶ 6. This approach to redistricting allowed all incumbents’ existing districts to remain intact and meant that the at-large candidate had to campaign across and represent the entire state. Also in the 1960’s, Texas created the now infamous District 6—often known as “Tiger” Teague’s district—which ran from Fort Bend County through rural east Texas into the southern ends of both Tarrant and Dallas Counties. See State Exh. 41.
The 1971 round of Congressional redistricting was notable at least in part because of the great lengths to which the state legislature went to solicit the views of incumbent congressmen. The Senate Congressional Redistricting Subcommittee actually flew to Washington to meet with the Texas delega*1313tion as a group and on an individual basis.9 See State Exh. 23, ¶8. In the 1980’s, the Texas Legislature managed to put together a plan despite two novel facts—the first Republican governor elected in Texas since Reconstruction and the applicability of § 5 of the Voting Rights Act to Texas Congressional redistrieting. See United States Exh. 1071 at 14; Plaintiff Exh. 28A (Map of 1980’s Plan C001).
C. The 1991 Congressional Redistricting Process
1. General Background
The Texas Constitution requires the Texas Legislature to redraw Congressional Districts after each Decennial Census. See Tex. Const, art. Ill, § 26. The Texas Legislature is a bicameral body consisting of the Senate and the House of Representatives. See Stip. 4. In 1991, the Texas Senate had 31 members, elected from single member districts. See Stip. 5. Of the 31 members of the 1991 Senate, 22 were Democrats and nine were Republicans, two were African-American, five were Hispanic and 24 were Anglo. All of the African-American and Hispanic members were Democrats. See State Exh. 1. The Texas House of Representatives had 150 members, also elected from single-member districts. See Stip. 6. Of the 150 members of the 1991 Texas House, 93 were Democrats and 57 were Republicans, 13 were African-Americans, 20 were Hispanic, and 117 were Anglo. As in the Senate, all of the African-American and Hispanic House members were Democrats. See State Exh. 2.
The following committees and subcommittees of the Texas Legislature were involved in the task of redistrieting in 1990 and 1991: the Senate Select Committee on Legislative Redistrieting, chaired by Senator Bob Glasgow; the House Committee on Redistrieting, chaired by Representative Tom Uher; the Senate Committee of the Whole on Redistricting, chaired by Senator Chet Brooks, which had two subcommittees—the Subcommittee on Congressional Districts, chaired by then-State Senator Eddie Bernice Johnson, and the Subcommittee on Legislative Redistricting chaired by Senator Bob Glasgow;10 and the Senate Committee of the Whole, chaired by Senator Chet Brooks.
The Senate Select Committee on Legislative Redistrieting and the House Redistricting Committee held joint regional outreach hearings throughout the state. Specifically, the committee heard and received testimony from individuals and organizations concerned about redistrieting in the following cities: Austin (February 28,1990); Lubbock (March 16,1990); Amarillo (March 17,1990); Corpus Christi (April 6, 1990); El Paso (May 18, 1990); Midland/Odessa (May 19, 1990); Houston (June 1, 1990); Beaumont (June 22, 1990);' Tyler (June 23, 1990); Fort Worth (July 13,1990); Dallas (July 14,1990); Laredo (July 27,1990); Edinburg/Harlingen (July 28, 1990); San Antonio (August 25, 1990); and Austin (September 28, 1990). In 1991, the Senate Committee of the Whole on Redistricting held its own public outreach hearings on a more limited scale in Houston (April 5, 1991), Brownsville (April 26, 1991), and San Antonio (May 1, 1991). See United States Exh. 1086.
*1314The transcripts and/or summaries of these numerous regional hearings are voluminous. What role these hearings ultimately played in Congressional redistrieting is difficult to ascertain. At least one Texas House member, Representative Kent Grusendorf, testified that the regional outreach hearings— often ill-attended by legislators—“essentially had no effect on the outcome of the redistricting process.” 6/27/94 TR. at 96. Grusendorf further observed: “At the time I think I thought it was serious, but in hindsight I think for the most part it was show.” Id. Nevertheless, the state relied strongly on citizen participation in these hearings when it justified its plan to the United States Department of Justice.
The Texas Legislative Council advised the Texas Legislature on legal issues of concern in drafting Congressional redistrieting legislation. See Dep. of Archer at 7. Jeff Archer, lead lawyer for the Council on redistricting, made presentations to committee members at the regional public outreach hearings on various legal issues to be considered in the redistrieting process. See id. at 87-88. In addition, the Council published a series on redistrieting—dubbed the “gray books”— that together served as a more comprehensive statement of state and federal law applicable to redistrieting. See Plaintiff Exhs. 13A, 13B, 13C. The Legislative Council also developed and had jurisdiction over REDAPPL (a/k/a “Red Apple”), which was the primary software used in drawing maps during the Congressional redistrieting process. See Dep. of Archer at 30.
The challenged redistrieting plan (HB1) generated litigation before it was even passed. On May 24, 1991, Republican plaintiffs in Terrazas v. Slagle, 821 F.Supp. 1162 (W.D.Tex.1993), brought an action under the Fourteenth and Fifteenth Amendments to the Constitution and the Voting Rights Act against various officials of the State of Texas and the Texas Democratic Party. In their First Amended Original Complaint, filed after the adoption of HB1, the plaintiffs in
Terrazas challenged the 1991 Texas Congressional Redistrieting Plan as unconstitutional and violative of the Voting Rights Act and alleged that it
sacrifices the rights of racial and political minorities to enhance the reelection chances of Anglo Democrat incumbents by fragmenting and concentrating the population centers of Hispanics and Republicans, diminishing the likelihood that candidates of their choice can be elected from within their communities.
United States Exh. 1005. The court in Terrazas ruled that the 1991 Texas Congressional Redistrieting Plan did not dilute the voting rights of racial, ethnic, or political minorities in violation of the Constitution or the Voting Rights Act. See Terrazas v. Slagle, 821 F.Supp. 1162 (W.D.Tex.1993).
The implementation of the challenged plan did increase the minority composition of the Texas Congressional Delegation. During the consideration of redistrieting by the Texas Legislature in 1991, the Texas Congressional Delegation had 27 members, of whom 18 were Democrats, nine were Republicans, one was African-American, four were Hispanic, and 22 were Anglo. See Stips. 46-47. As a result of the 1990 Census, the Texas Congressional Delegation increased to 30 members. See Stip. 48. Of the 30 members of the Texas Congressional Delegation elected in 1992 after the 1991 redistrieting—20 Democrats and ten Republicans—two are African-American, five are Hispanic, and 23 are Anglo. See Stips. 49-50.
2. Voting Rights Act Considerations
The Legislature embarked upon Congressional redistrieting against the legal backdrop of the Voting Rights Act. As described swpra, the Texas Legislative Council through the “gray books” attempted to summarize Voting Rights Act concerns for the Texas Legislature.11 Further, Jeff Archer, the *1315Council’s lead lawyer on redistricting, frequently testified before the Legislature’s committees on the Voting Rights Act requirements. For instance, Archer told the Senate Committee of the Whole on Redistricting that “mere lack of proportional representation is not enough” to establish a violation of the Voting Rights Act, but is “strong evidence.” Plaintiff Exh. 16 at 10.
The Legislature also heard from concerned citizens and organizations about the Voting Rights Act considerations relevant to Congressional redistricting. For example, George Korbel, director of litigation for Texas Rural Legal Aid and former regional director of the Mexiean-Ameriean Legal Defense Fund, testified before the Senate Committee of the Whole on Redistricting that “unless there is [sic] at least two additional Hispanic Congressional districts and one additional Black Congressional district, ... the reapportionment of the Congress is not going to pass the Department of Justice.” United States Exh. 1086 (4/5/91).
Once the redistricting legislation had passed, the Voting Rights Act considerations in HB1 were set forth in a September 1991 attachment to the State’s § 5 submission entitled Narrative of Voting Rights Act Considerations in Affected Districts prepared by the Texas Congressional Redistricting Staff. See Plaintiff Exh. 4C. As the document sets forth in its introduction, the Narrative functions to “give an overview of the efforts made to address Voting Rights Act concerns.” Id. at 1.
The Narrative begins by noting the legislative agreement that the three new Congressional seats apportioned to Texas
should be configured in such a way as to allow members of racial, ethnic, and language minorities to elect Congressional representatives. Accordingly, the three new districts include a predominantly black district drawn in the Dallas County area and predominantly Hispanic districts in the Harris County area and in the South Texas region. In addition to creating the three new minority districts, the proposed Congressional redistricting plan increases the black voting strength of the current District 18 (Harris County) by increasing the population to assure that the black community may continue to elect a candidate of its choice.
Id. After making these initial observations, the Narrative analyzes the three new minority districts as well as District 18 in greater detail.
The Texas Legislature agreed that a new “safe” African-American district should be drawn in Dallas County.12 See id. at 2. The African-American community in Dallas County insisted on a 50% total African-American population for the district “which the community felt was necessary to assure its ability to elect its own Congressional representative without having to form coalitions with other minority groups.” Id. Meeting the threshold 50% figure meant that more compact alternative proposals for District 30 had to be rejected.13 See id.
Texas legislators were aware that the failure to draw an African-American majority district in Harris County in 1991 might be interpreted as retrogression under Section 5 of the Voting Rights Act. See 6/30/94 TR. at 4-31; United States Exh. 1047. Therefore, in order to keep District 18 as a “safe” African-American district, “additional black population was taken from adjacent districts thereby increasing the total African-Ameri*1316can population to 50.9% and decreasing the total Hispanic population to 15.3%.”14 Plaintiff Exh. 4C at 5. The remaining Hispanic population was placed in District 29—the new “safe” Hispanic district—consisting of a 60.6% Hispanic population and a 10.2% African-American population. See id. The Narrative concludes that the changes in District 18 and the configuration of District 29 “result in the maximization of minority voting strength for this geographical area.” Id.
In the heavily Hispanic South Texas area, the Legislature faced no major problems concerning minority voting strength or adjustments of population totals. See id. District 28—the new “safe” Hispanic district in South Texas—was drawn with the constant input of the minority leadership in Bexar County and the Rio Grande Valley. See id. The location of District 28 in the northern portion of South Texas was determined in part by the historical north-south configuration of Congressional Districts 15 and 27. This was the result of attempts to remedy a January 29, 1982 Section 5 objection which expressed concern that the original east-west configuration of Districts 15 and 27 resulted in packing of the Hispanic population. See United States Exh. 1065 at 9; 6/29/94 TR. at 3-169.
a. Racial Polarization
In configuring Congressional District 30 in Dallas County, the African-American community sought 50% total African-American population as the minimum necessary to assure that a candidate of choice would be elected. The 50% figure was deemed significant because “[tjhere is little evidence of coalition voting between blacks and Hispanics in Dallas County” and reaching the threshold 50% would obviate any need to form coalitions. Plaintiff Exh. 4C at 3.
In Harris County, University of Houston political scientist Dr. Richard Murray observed that
the political alliance that had been forged between blacks and Hispanics in the 1960’s began to break down. Open electoral conflicts became more common. Relations were especially strained in 1989 when, in a contest for an open at-large seat on Houston’s city council, and [sic] African American, Sheila Jackson Lee, upset the favored Hispanic, former city controller Leonel Castillo. Hispanics returned the favor in 1991 when Grade Guzman Saenz unseated a black incumbent in another at-large election. And in a hard-fought mayoral race in 1991, African Americans rallied behind black Texas Representative Sylvester Turner, given [sic] him 97% of their votes in a runoff. Hispanics supported the Anglo winner, Bob Lanier, by a nearly three to one margin.
Lawson Exh. 26 at 15. This breakdown of past coalitions prompted Hispanic strategists to argue that Hispanics and African-Americans should not be combined in a new Harris County Congressional district. See id. As Houston City Councilman Ben Reyes testified at an outreach hearing held in Houston, combining minority groups in nearly equal numbers in a new Harris County district would be the “worst scenario” because “they will vote for members of their own ethnic group, making it more likely that a non-minority candidate will win.”15 Plaintiff Exh. 15H at 18.
In general, some racial or ethnic polarization occurs in majority-minority districts in Texas. See Plaintiff Exh. 36 at 23-27. The *1317analysis of Dr. Allan J. Lichtman, an expert for the State of Texas, concluded that Anglos usually bloc-voted against Hispanic candidates in the majority-Hispanic districts. In each of four categories, a mean of 21% or fewer Anglo voters voted for Hispanic candidates. See State Exh. 14 at 21 (Table 5). For the African-American majority districts, Dr. Lichtman similarly concluded that Anglos usually bloc voted against African-American candidates. In each of four categories, a mean of 34% or fewer Anglo voters voted for African-American candidates. In all categories but the legislative (which included only one election for each district), a mean of 25% or fewer Anglo voters voted for African-American candidates. See id at 22. Dr. Ronald Weber, an expert for the plaintiffs, conceded some racial or ethnic polarization, but concluded that it is “not legally or politically consequential.” Plaintiff Exh. 36 at 27.
b. History of Discrimination
Texas has a long, well-documented history of discrimination that has touched upon the rights of African-Americans and Hispanics to register, to vote, or to participate otherwise in the electoral process. Devices such as the poll tax, an all-white primary system, and restrictive voter registration time periods are an unfortunate part of this State’s minority voting rights history. See United States Exh. 1065 at 3; State Exh. 17 at 6. The history of official discrimination in the Texas election process—stretching back to Reconstruction—led to the inclusion of the State as a covered jurisdiction under Section 5 in the 1975 amendments to the Voting Rights Act. Since Texas became a covered jurisdiction, the Department of Justice has frequently interposed objections against the State and its subdivisions. See United States Exh. 1095.
3. Incumbents’ Interests
As has historically been the case, Congressional incumbents were actively involved in the redistricting process. The Texas Democratic Congressional Delegation formed a redistricting committee which began work in late 1989 or early 1990. See Lawson Exh. 3, ¶4. Congressman Ron Coleman, a Demoerat from El Paso and head of the Redistricting Committee, asked Democratic incumbents what areas they wanted to represent; to the extent their preferences overlapped, Coleman mediated between incumbents. See id The committee’s “overriding objective,” however, was incumbency protection. Id at ¶10.
The Delegation played a significant role in determining the configuration of the Congressional districts, developing their own alternative plans and presenting those plans to state legislators. See Lawson Exh. 3, ¶ 9. As Congressman Coleman observed: “We drew our own plans and presented them to the Legislators, and various members of the Delegation met with [Ljegislators in Austin to discuss the incumbents’ needs and preferences. The Delegation was definitely a force in the process.” Id
Not surprisingly, Republican incumbents were active in the redistricting process as well. For instance, Congressman Joe Barton urged the joint redistricting committees to protect as many incumbent congressmen as possible. See Plaintiff Exh. 15H at 25. In short, as a general rule, incumbents sought to influence the Legislature to draw districts that would maximize their chances for reelection. Furthermore, members of the Legislature openly acknowledged the role of incumbents on the redistricting process. See United States Exh. 1092 (Senate Committee of the Whole, transcript dated 8/24/91, at 17) (statement of Senator Eddie Bernice Johnson: The incumbents “have practically drawn their own districts. Not practically, they have.”); Plaintiff Exh. 23 at 21 (statement of Representative Uher: “Well, I think that not just a congressman but a large majority of the Congressional delegation have endorsed the basic plan that we started with, and that’s the reason that we’re still adhering to that basic plan, is because of a majority of support of the Congressional delegation and not just necessarily one individual congressman’s support.”).
Incumbency protection would by definition at minimum require that incumbents not be paired against each other. Plan C657 reflects a successful effort to avoid this result. See Stip. 52. As shown by the various maps *1318making up State Exh. 9B, incumbent residences repeatedly fall just along district lines. Congressman Lamar Smith’s residence lies in an inlet in Bexar County in the last voter tabulation district (“VTD”) before his Congressional District 21 ends at the northern boundary of Congressman Henry B. Gonzales’ District 20.16 Congressman Gonzales’ residence in turn lies in the last VTD before his district ends at the southern boundary for District 21. In Dallas County, Congressman Frost’s residence lies in a small indentation jutting into a part of District 30, and Congressman Bryant’s residence lies just barely on the other side of a dividing line between District 5 and District 30. Finally, in Harris County, small indentations permit Congressman Jack Fields’ residence to remain in District 8 and Congressman Andrews’ residence to remain in District 25.
At least as measured by 1992 election results, the incumbents experienced great success in redistricting to assure incumbency protection. Each incumbent member of Congress—except Albert Bustamante, an Hispanic Democrat who represented District 23—was reelected to Congress in 1992. Bustamante was defeated in the 1992 General Election by Henry Bonilla, an Hispanic Republican. See Stip. 51. Additional specific instances of incumbent interests will be detañed infra in the analysis of particular Congressional Districts.
4. Use of Racial Data
As with incumbency interests, particular instances in which racial data were used in redistricting are hereafter detaüed in discussing individual Congressional Districts and their shapes. However, some general observations about racial/ethnie information widely avaüable to legislators are appropriate.
Redistricting data were available to all members of the Legislature and their staffs as well as to any groups and individuals sponsored by members of the Legislature. See Plaintiff Exh. 13D at 1. The primary software used for map drawing in Congressional redistricting—REDAPPL—was readily avaüable on work stations in the redistricting offices of the Texas Legislative Councü. See id.
The feature of REDAPPL of most interest is the system’s ability to provide racial and ethnic data at both the VTD and block level. REDAPPL software allowed the operator to work at the VTD level and call up racial/ethnic information in addition to other types of information such as population, voting age population, incumbent location, and street names. See 6/29/94 TR. at 173. Election contest information was avaüable at the VTD level, see id., but REDAPPL did not aüow the operator to work with multiple elections simultaneously on a screen.17 Indices of partisanship such as the NCEC Index, prepared at a national level for the use of Democrats, involved multiple elections, but they could not be accessed on a REDAPPL screen. See id. at 175.
The critical feature of REDAPPL is that it allowed the operator to “split” a VTD and work on a block-by-block level. See id. at 176. Racial/ethnie breakdown was avaüable on a block level on REDAPPL. See id. at 177. By contrast, no election contest information was avaüable at the block level on the REDAPPL software.18 See id. In sum, REDAPPL aUowed the user to work with racial/ethnic data at even the block level; election information was simply unavaüable at that level through REDAPPL.
If the Legislature intended to allocate voters on the basis of race, REDAPPL certainly provided a readüy avaüable, efficient means of doing so. In fact, because the software constantly displayed racial and ethnic data on the screen anytime an operator used the system, a would-be map drawer would affir*1319matively have to ignore the data. See 6/30/94 TR. at 81. But as Chris Sharman, the principal computer technician/map drawer involved in Congressional redistricting, testified:
The problem is when you draw on this computer, it tells you the population data, racial data. Every time you make a move, it tabulates right there on the screen. You can’t ignore it.
Id.
5. Congressional District 30
Describing the boundaries of Congressional District 30 is no easy task. Even the State of Texas has difficulty analyzing the contours of this extraordinarily oddly-configured, sprawling district. See Appendix (Map of District 30). Dan Weiser, an expert employed by the United States, prepared a map isolating Congressional District 30 and breaking it down into a “core” and no less than seven segmented portions. See State Exh. 33. Even under the State’s analysis, the “core” of District 30 accounts for only 50% of the district’s voting age population. See id.
The “core” of the district includes what is generally known as South Dallas, Fair Park, and portions of South Oak Cliff and Pleasant Grove in southeast Dallas. See United States Exh. 1070 at 9 (declaration of Dr. Paul Waddell). The district then moves northeast from its core and splits into a northern and a western extremity. See id. at 10. The western extremity proceeds to incorporate much of West Dallas before it branches off to the north and south. The southern portion gathers the “older core” of Grand Prairie, while the northern portion moves “through mostly undeveloped land between Irving and Arlington, and into the DFW Airport.” Id. at 11. The northern extremity includes a portion of the district to the northeast of the Central Expressway characterized as “particularly complex” by the State’s land use expert, Dr. Paul Waddell. Id. at 12.
As even a cursory glance at the map of District 30 in isolation reveals, the district can really only be described here in the most general terms as its meanderings are too complicated and frequent to detail. See State Exh. 33. Thus, the remainder of the discussion of Congressional District 30 in this portion of the opinion will focus exclusively on the intent of the Texas Legislature in drawing the convoluted boundaries of District 30.
According to the Narrative of Voting Rights Act Considerations in Affected Districts, an attachment to the State’s § 5 submission, the Texas Legislature agreed that a “safe” African-American Congressional District would be drawn in Dallas County.19 See Plaintiff Exh. 4C at 1-2. The African-American community insisted upon a 50% total African-American population in order to assure a “safe” African-American district in Dallas County. See id. at 2. The community succeeded, as Congressional District 30 configured under Plan C657 had a total African-American population of 50.0% and a total Hispanic population of 17.1%. See id.
The trial testimony of District 30 Congresswoman Eddie Bernice Johnson in Terrazas v. Slagle—a previous challenge to C657 under the Constitution and Voting Rights Act—is consistent with the plainly stated conclusion in the Narrative that the Texas Legislature intended to create a safe African-American district in Dallas County. In response to a question about whether a dominant goal existed in redistricting Congressional Districts in Dallas County, Johnson— who at the time of redistricting chaired the Senate Subcommittee on Congressional Districts and was a representative of Dallas County in the Texas Senate—replied:
Yes. I had made a commitment to that Black community, that they would have a safe district, as had been mandated and *1320expected for a number of years, and I did not intend to go home without that.
Plaintiff Exh. 8B at 231.
In explaining the boundaries of the district, Congresswoman Johnson testified that the district was drawn this way as a result of two competing tensions—namely that the district was intended to be a “safe” African-American district and the African-American population in Dallas County had over time dispersed from its previous “core” location. The following exchange is particularly informative:
Q: When you say they deteriorated, what do you mean in that respect, Mrs. Johnson?
A: Well, the population had moved— started to move out; there were lot of boarded up houses. The whole core of that area was moving out. There had been a deterioration of about 40 to 45 to 50 percent in certain areas of voters in a 10 year period. In addition to that, there were a large number of persons there who were felons, who could not vote. So, though they were over 18, it substantially deteriorated their voting strength. We then attempted to locate where that population shifted to. And in attempting to trail that—to trace that population, we could see that it was moving outer and around. It was going into the Grand Prairie area and into the Pleasant Grove area. And then there were pockets of persons who had lived here, and then this was moved—
Q: Who lived in the Black core district?
A: —who lived in the core district, and also in the north end of Dallas County, into Collin County. Those were performing voters who expressed a desire to be in the minority district____
Q: Well, to the extent then that we see fingers of the ... district going off in the north—north Dallas County, and even into southern Collin County, I suppose, we are talking about these are Black migration areas that you were attempting to bring into the district; is that correct?
A: That is correct.
Id. at 233-35. In sum, Congresswoman Johnson testified in Terrazas that the shape of Congressional District 30—including the various “finger”-like extensions that are common northeast of the Central Expressway—• can be understood as a conscious effort to pick up African-American voters who had dispersed from the core area.20
When asked about the influence of incumbent Congressmen Martin Frost and John Bryant on the shape of District 30, Congresswoman Johnson testified that her sole focus in drawing the district was on looking out for African-American voters:
Q: All right. Now, was anything done in the course of the creation of this map, Mrs. Johnson, that you could tell us about, to aid Congressman Bryant or Congressman Frost? Or did this map just happen this way?
A: I got beat up so many times because I wouldn’t do anything but look out for Black voters.
Id. at 247. Johnson—the principal architect of District 30—proceeded to testify that in drawing the district she was able to pick and choose the “performing” African-American voters she wanted to include, leaving the “nonperforming” African-American voters for Bryant and Frost.21 See id. at 248.
*1321Representative Fred Blair’s testimony at trial in Terrazas is consistent with Congresswoman Johnson’s view expressed in that same ease that the shape of the district can be explained as an effort to locate and select “performing” African-American voters in order to guarantee the African-American community a safe African-American seat. Blair—a Texas House member from Dallas— observed that District 30
was crafted in a manner that we sought to pick up those precincts, those communities, those areas that we thought were stable areas that would present an opportunity to elect an African-American____ [I]n looking at developing a Congressional Plan, we wanted to find those areas that we thought were stable areas. Homeowners were very important to us. We wanted to make sure we included a significant number of those within a district because just to lump in African-Americans and say we have an African-American district that may have a number of apartments where there is a lot of movement going on, we thought we had to be very sure, very careful, in drawing lines so that we could create a district that we thought was winnable with a 50 percent.
Plaintiff Exh. 8D at 108-09. Furthermore, his testimony directly linked this effort to find “stable areas” with the irregular shape of the district. See id. at 109.
The testimony submitted in this racial gerrymandering case is at first glance starkly at odds with the explanation for the district’s severely contorted boundaries offered in Terrazas, which was of course not a racial gerrymandering case. The most prominent example is the testimony offered by Congresswoman Johnson in this case.
Unlike Terrazas, where she did not acknowledge that Congressmen Frost and Bryant had a role in determining the district’s boundaries, Congresswoman Johnson testified for purposes of this case that District 30 did not include some portions of the area encompassed by her senate district because the incumbent congressman in District 5—John Bryant—wanted the area: “[Five] wanted voters that they had previously represented, just as 24, just as 6, just as 5, just as everybody that had a stake in it. Everybody wanted as much of what they had previously represented as possible on both sides of the political spectrum.” Dep. of Johnson at 82. Johnson further testified that a more compact African-American majority district could have been drawn in the Dallas area if she did not have to address the concerns of incumbents.22 See Dep. of Johnson at ISO-32, 142.
Other testimony in this case emphasized the role of incumbents in the shaping of District 30’s bizarre boundaries. Ted Lyon, a former member of the Texas House and Senate involved in Congressional redistricting in 1980 and again in 1990, described in general terms the active role of incumbents in drawing District 30:
In focusing on the incumbent congressmen in the Dallas area, it became clear almost immediately that there would be a fight between Congressmen Frost and Bryant, on the one hand, and Senator Eddie Bernice Johnson, on the other, over the African-American voters who had previously resided in Districts 3 [sic] and 5. Frost and Bryant were not concerned about the race of these voters. They just wanted to hold onto enough Democrats to assure reelection. Senator Johnson was trying to take both minority and Democratic voters from what had previously been Districts 24 and 5 in order to construct a majority-black district that would satisfy the Voting Rights Act. Conflict arose, of course, because Democratic populations and African-American populations are often the same. The redistricting process became a no holds barred political fight, and fangs were out.
Lawson Exh. 14, ¶8.
Lyon also testified about the specific impacts of incumbency protection on the contours of the district. For instance, he attributed the “irregular” shape of District 30 in Oak Cliff and Grand Prairie to fighting be*1322tween Frost and Johnson eventually settled by essentially splitting the areas between Districts 24 and 30. See id. at ¶ 9. Lyon also attributed in part some of the irregularity in the district’s eastern shape to incumbency protection—namely keeping Congressman Bryant’s East Dallas neighborhood in District 5.23 See id. at ¶ 11.
Apparently, Democratic incumbents in Dallas County were quite interested in keeping African-American voters in their newly configured districts. Congresswoman Johnson testified that Congressman Frost was “looking for voters [in the urban areas of Dallas county] that were going to vote in the Democratic Primary, and clearly he was more likely to be sure of it if they were black.” Dep. of Johnson at 129-130. Quite telling is an August 28,1991 letter written by then-Senator Johnson to John Dunne, then the head of the Civil Rights Division at the Department of Justice, in which the Senator requested a review of proposed Districts 12 and 24 and their potentially dilutive effect on the minority community in Tarrant County. See Plaintiff Exh. 6E6. In the first paragraph of this letter, Senator Johnson explains why African-American voters were so attractive to incumbents fighting over district boundaries:
For primary elections, approximately 97% of the total votes cast by Blacks in the Dallas/Fort Worth metroplex area are cast in the Democratic primary. Because of the consistency of this voting pattern, Democratic incumbents generally seek to include as many Blacks as possible into their respective districts. Throughout the course of the Congressional redistricting process, the lines were continuously reconfigured to assist in protecting the Democratic incumbents in the Dallas/Fort Worth metroplex area by spreading the Black population to increase the Democratic party index in those areas.
Id, In other words, incumbent protection in Dallas County involved the allocation of African-American voters among the districts.
Again in contrast to her prior Terrazas testimony, Congresswoman Johnson described in her deposition in this case a variety of nonraeial factors that went into the drawing of District 30’s boundaries. For instance, she testified that the district mappers “made an effort to put communities of interest together in this district. We made an effort to identify voters that would support the same kinds of major issues in the same manner, notwithstanding their color.” Dep. of Johnson at 32. Johnson further asserted that she and her staff considered the result of certain votes in deciding whom to include in District 30:
We looked at a couple of referenda votes for the Dallas area rapid transit system. We also looked at a bond election vote for the Dallas independent school district trying to determine where there might be more communities of interest, where there would be support that would go beyond the color of the candidate.
Dep. of Johnson at 144.
Other testimony ostensibly supports Congresswoman Johnson’s suggestion that “communities of interest” were put together in District 30. In a report dated June 24, 1994, Dr. Paul Geisel, an expert for the state, proclaims that District 30 represents a community of interest that shares “one economy, one transportation system, one media/communieations system and one higher educational system.” State Exh. 18 at 7.
Dr. Paul Waddell, the United States’ land use expert, in a report prepared June 21, 1994, attempts to explain the boundaries of District 30 in nonraeial land use terms. *1323First, most of the “arms and fingers” of Congressional District 30 follow both natural and commercial land use boundaries, including industrial belts, retail areas, the Trinity River, and freeway corridors. See United States Exh. 1070 at 8. Second, District 30’s extremities encompass within their boundaries little single-family land use, but “clusters of multi-family land use.” Id. Third, the extremities of the district incorporate “substantial land use areas that are office, industrial, retail, or airport land use areas, even when these areas do not clearly serve as a bridge to other residential areas.” Id. Fourth, the district boundaries do not, upon close analysis, divide single family residential neighborhoods, but in fact encompass “multifamily areas and avoid established single family neighborhoods.” Id.
Neither Dr. Waddell nor Dr. Geisel suggested that the Legislature had these particular “communities of interest” in mind when drawing the boundaries of District 30. While Fred Blair’s testimony in Terrazas suggests that the map drawers preferred to include home-dwellers over apartment-dwellers, an assertion at odds with Dr. Waddell’s conclusions, the record is otherwise void of support for any land use thesis. In sum, both reports undoubtedly accurately describe the district, but are more properly seen as post hoc descriptions of the boundaries.
6. Congressional Districts 18 and 29
According to the Narrative of Voting Rights Act Considerations in Affected Districts, the Legislature sought to create a “safe” Hispanic seat in the new Harris County Congressional District 29 as well as increase African-American voting strength in District 18 in order to assure that the African-American community could continue to elect a “candidate of its choice.” Plaintiff Exh. 4C at 1. Prior to redistricting in 1991, District 18 was underpopulated by 116,549 people and was made up of 35.1% total African-American population and 42.2% total Hispanic population.24 See id. at 5. To “remedy” this situation, additional African-American population was taken from adjacent districts, thereby increasing the total African-American population to 50.9%. The remaining Hispanic population was shifted over to the new Hispanic District thereby decreasing the total Hispanic population in District 18 to 15.3%. See id. For its part, District 29 consists of 60.6% total Hispanic and 10.2% total African-American population. See id.
An appreciation for the precision with which this segregation of Hispanies and African-Americans in Harris County was carried out may not be had without a detailed look at the map of District 18 based on African-American population distribution by Census block and the map of District 29 based on Hispanic population distribution by Census block. See Plaintiff Exh. 55 and 53. The detail allowed by these maps highlights in District 18, for example, the “many narrow corridors, wings, or fingers that reach out to enclose black voters, while excluding Hispanic residents.” Richard H. Pildes & Richard G. Niemi, Expressive Harms, “Bizarre Districts, ” and Voting Rights: Evaluating Election-District Appearances After Shaiv v. Reno, 92 Mich.L.Rev. 483, 556 (1993) (hereinafter, “Pildes & Niemi”). District 29’s border is similarly characterized by fingers reaching out to enclose Hispanies. In fact, these districts are so finely “crafted” that one cannot visualize their exact boundaries ■without looking at a map at least three feet square.25
The geographic dispersion of the various minority communities within Harris County is definitely an obstacle in drawing a majority-minority district each for African-Americans and Hispanies. As Dr. Ronald Weber, the main expert for the plaintiffs, testified at trial, “[T]he Hispanic community is dispersed in two quadrants about 10:00 or 11:00 o’clock on [Plaintiff Exh. 53] ... and about 5:00 o’clock” while the African-American commu*1324nity has three centers of concentration, “one as [sic] approximately 10:00 o’clock [on Plaintiff Exh. 55], which is to the west of the Hispanic concentration, one at about 1:00 o’clock, which is another African American concentration, and then finally at about 6:00 and 7:00 o’clock.” 6/28/94 TR. at 263. This dispersion helps account for the fact that District 29 cuts through the center of Houston to join the two Hispanic quadrants and that District 18 snakes around the city to capture the various African-American concentrations. See id. at 264; Appendix (Maps of Districts 18 and 29); see also Plaintiff Exhs. 34H8, 34H9 (small maps).
In the earliest stages of the Congressional redistricting process, state Democratic and Republican leaders rallied behind the idea of creating a new Hispanic safe seat in Harris County while preserving the safe African-American seat in District 18.26 Also early on in the redistricting process, Texas House member Roman Martinez, an Hispanic Democrat from Houston, announced a plan for drawing the new Hispanic district while maintaining majority-minority District 18 and preserving the “Democratic nature” of Congressional District 25. See Plaintiff Exh. 9. Representative Martinez—one of two Hispanic members of the Texas House from Harris County—would play a major role in the drawing of District 29, as would another Congressional aspirant, then-State Senator Gene Green.27 In the press release announcing his plan, Martinez promised that
what Houston’s Hispanic community has long worked for—its own Congressional district—will be accomplished. My hope is that it will be accomplished through the legislative process as we present this plan to both the House and the Senate in the State Legislature. But if this district as we envision it is not a product of the legislative process, we will enlist the help of the U.S. Department of Justice and the courts under the jurisdiction of the Voting Rights Act. ■
Id.
Representative Martinez’s testimony at trial in Terrazas v. Slagle is consistent with the analysis offered in the Narrative on the creation of the two majority-minority districts in Harris County. As the primary architect of the lines in Harris County, Martinez defined his goals in redistricting:
Again, the first goal was to assure no retrogression for the 18th Congressional District, insuring that that was maintained as an African-American district. And then, secondly, it was a very important goal as a Hispanic representative to insure that we created for the first time a Congressional seat for the Hispanic community to elect the first Hispanic Congressman.
Plaintiff Exh. 8A at 134, 146. These “goals” took hold in the Legislature. For example, Carl Reynolds, a staff member of the Senate Subcommittee on Legislative Redistricting who was actively involved in Congressional redistricting, observed that “[t]here was an understanding that that new district [in Harris County] would be a—I think a 60 percent or more Hispanic district.” Dep. of Reynolds at 62.
At least two other factors influenced the boundary drawing of the Harris County districts.28 First, to the extent possible, Con*1325gressman Mike Andrews’ district—Congressional District 25—was to be kept intact and Democratic. See Dep. of Martinez at 21. Ed Martin, Executive Director of the Texas Democratic Party, testified that a suggestion by Congressman Craig Washington that District 18 be reconfigured based on the shape of African-American majority Senate District 18 was unacceptable “because it would have taken a large chunk out of District 25.”29 Lawson Exh. 15, ¶ 18. Congressional District 25 was not spared in the redistricting process, however, as it eventually lost population to District 18 in southern Harris County thereby necessitating the additional population from the southwest, the Baytown area, and north of the Ship Channel. See 6/30/94 TR. at 445.
The second factor was the desire of Senator Gene Green to draw a Congressional district in which he could run, namely one which included as much of his Senate district as possible. See Dep. of Martinez at 22. Not surprisingly, Senator Green was not entirely successful in influencing the map drawing. For instance, an alternative to the present configuration of District 29 would have brought District 29 around the northeast portion of Harris County. See State Exh. 15 at 4. This alternative included areas perceived as more favorable to Senator Green. See id. As per Representative Martinez’s wishes, the downtown links the two major quadrants of District 29 in its final form.30
Senator Green’s insistence on including his primarily Anglo “home” precincts in District 29 had the effect of diluting the Hispanic population percentage of the district. See Dep. of Martinez at 28. Representative Martinez testified that it became necessary to operate at the Census block level, in order to reach the agreed upon 61% Hispanic percentage. See id. In sum, the inclusion of certain Anglo precincts in District 29 necessitated operating at the block level—that is, necessarily splitting precincts—to find Hispanic voters to meet the target percentage.31 See id.
The effect of splitting dozens of VTD’s to create Districts 18 and 29 was an electoral nightmare. Harris County estimated that it must increase its number of precincts from 672 to 1,225 to accommodate the new Congressional boundaries. Polling places, ballot forms, and the number of election employees correspondingly multiplied. Voters were thrust into new and unfamiliar precinct alignments, a few with populations as low as 20 voters.32
7. Congressional District 28
The enormous growth in the Hispanic population in Texas—particularly South Texas— assured that a new Hispanic seat would be drawn in the South Texas area. See Plaintiff Exh. 4C at 5. According to the Narrative of Voting Rights Act Considerations in Affected Districts, construction of the South Texas district posed no major problems concerning minority voting strength or adjustment of population numbers/pereentages because of the heavy Hispanic concentration throughout the region. See id. at 5-6. District 28 as constructed runs from Bexar County south to Starr and Zapata Counties and is 60.4% Hispanic in total population and 8.5% African-American.33
*1326Then-Senator Frank Tejeda, Vice-Chair of the Senate Subcommittee on Congressional Districts, was primarily responsible for the development of the South Texas districts generally and in particular for the drawing of the new 28th District. See id. at 5; LULAC Exh. 6 at 5. As now Congressman Tejeda concedes in his affidavit submitted in this case, he attempted to draw a district which would facilitate his potential candidacy. See LULAC Exh. 6 at 8. To this end, Tejeda included south Bexar County and much of east San Antonio in District 28—both areas which he had represented in the Texas Senate. See id.
Congressman Tejeda also testified that he and his staff attempted to comply with the wishes of the Texas Democratic Congressional delegation in drawing districts in South Texas. See id. at 5. For example, Congressman Kika De La Garza insisted that Hidalgo County not be divided between districts, and Congressman Solomon Ortiz “let it be known that he did not want either Cameron or Nueces County split.” Id. at 6. None of these counties ended up split in the final redistricting plan. Keeping these counties whole meant, however, that Jim Wells County and Kleberg County—apparently less politically powerful—would eventually be split. See id. at 9. As for the splits in Guadalupe County, Tejeda maintained he had no reason to believe the cuts were made for anything other than political reasons.34
Nick Dauster worked for Senator Tejeda during the 1991 Congressional redistricting and was the lead Senate staff person on South Texas Congressional redistricting. See LULAC Exh. 1 at 5. He explained that District 28 needed additional population and “it was suggested that we include Democratic communities in Guadalupe and Comal Counties.” Id. at 15. These communities were heavily minority, but—according to Dauster—had much in common with the portions of San Antonio in District 28. See id.
Finally, Congressman Tejeda observed that District 28 as constituted represents many communities of interest. The district is composed primarily of low to middle income, blue collar, working class families. See id. at 2. Much of the district is involved in agriculture and related industries. See id. The district also has a very large veteran population of approximately 10%. See id. at 3.
8. Other Congressional Districts
Many of the remaining districts challenged by the plaintiffs share a troubling characteristic: counties within the district appear to be split on racial lines. Lubbock County is a prime example of this trait among these districts. While the part of Lubbock County that was split into Congressional District 13—occupied by Democrat Bill Sarpalius— contained 77.4% African-American and Hispanic population, the part split into Republican Larry Combest’s District 19 contained only 19.2% African-American and Hispanic population. See Plaintiff Exh. 36H. During the Texas Senate floor debate on August 24, 1991, the day Plan C657 was passed by the Senate, then-Senator Eddie- Bernice Johnson, Chair of the Subcommittee on Congressional Districts, succinctly explained the split in Lubbock County:
BIVINS: But I’m referring to the 19th and the 13th Congressional districts specifically in pointing out that there are split counties and specifically—
JOHNSON: Yes, I can tell you why Lubbock is split. It splits to remove that Black community into a district where they can have more impact.
United States Exh. 1092 at 21.
Midland and Ector Counties raise the same issue. In Ector County, the part of the *1327county split into former Democratic Congressman Albert Bustamante’s District 28 contained 72.1% African-American and Hispanic population, while the part in Republican District 19 contained only 21.2% African-American and Hispanic population. See Plaintiff Exh. 36H. In neighboring Midland County, the part of the county split into Congressional District 23 contained 76.5% African-American and Hispanic population; the part split into District 19 contained less than 17% African-American and Hispanic population (16.2%); and the part split into Republican Lamar Smith’s District 21 contained only 10.9% African-American and Hispanic population. See id.
In explaining the Midland County split during floor debate on August 24, then-Senator Johnson maintained that “the Congressionals [i.e., the Congressional Delegation] feel that that minority population ought to go into that minority district and that’s where those minorities out there wanna go, and that’s the reason why they put them in there.” United States Exh. 1092 at 1 (tape 2). Johnson’s testimony in Terrazas explaining the Midland County split emphasized that the district lines were really responding to the desire of minorities to be in District 23:
We made sure we had the minorities’ input on these areas. I have heard comments about somebody in Midland, what have you. We were responding to minorities. They wanted to be certain places. They didn’t always feel free, they tell me, to even say that to anyone else, but they felt free to tell me. And we were responsive to them.
Plaintiff Exh. 8B at 243.
Speaking more generally, Congresswoman Johnson testified that county splits along minority lines in Plan C657 are not accidental. See Dep. of Johnson at 97. These county splits—whether reflecting the urgings of minorities or incumbents or both—occur throughout the challenged districts:
1.Nacogdoches County, Texas: While the part of Nacogdoches County that was split into Democrat Charles Wilson’s Congressional District 2 contained 52.5% African-American and Hispanic population, the part split into Democrat Jim Chapman’s District 1 contained only 13.2% African-American and Hispanic population. See Plaintiff Exh. 36H.
2. Montgomery County, Texas: The part of Montgomery County that was split into District 2 contained 40.9% African-American and Hispanic population, while the part split into District 8—represented by Republican Jack Fields—contained only 8.1% African-American and Hispanic population. See id.
3. Waller County, Texas: While the part of Waller County that was split into Democrat Greg Laughlin’s District 14 contained 56.9% African-American and Hispanic population, the part of the county allocated to District 8 contained only 8.6% African-American and Hispanic population. See id.
4. Brazos County, Texas: The part of Brazos County that was split into Democrat John Bryant’s District 5 contained 71.7% African-American and Hispanic population, while the part split into Congressional District 8 contained only 15.3% African-American and Hispanic population. See id.
5. Brazoria County, Texas: While the part of Brazoria County that was split into District 14 contained 32.0% African-American and Hispanic population, the part split into Republican Tom DeLay’s District 22 contained only 22.8% African-American and Hispanic Population. See id.
6. Hunt County, Texas: While the part of Hunt County split into District 1 contained 20.4% African-American and Hispanic population, the part of the county split into Democrat Ralph Hall's District 4 contained only 8.0% African-American and Hispanic population. See id.
7. Gregg County, Texas: The part of Gregg County that was split into District 1 contained 32.2% African-American and Hispanic population, while the part split into District 4 contained only 17.8% African-American and Hispanic population. See id.
8. Denton County, Texas: While the part of Denton County that was split into Bill Sarpalius’ District 13 contained 22.0% African-American and Hispanic population, the part split into Republican Joe Barton’s Dis*1328trict 6 and Democrat Hall’s District 4 contained only 10.0% and 9.2% African-American and Hispanic population respectively. See id.
9. Collin County, Texas: While the part of Collin County that was split into Congressional District 30 contained 57.4% African-American and Hispanic population, the part split into District 4 contained 17.2% African-American and Hispanic population and the part split into Republican Sam Hall’s District 3 contained only 7.7% African-American and Hispanic population. See id.
10. Smith County, Texas: The part of Smith County split into District 5 contained 64.0% African-American and Hispanic population, while the part of Smith County split into District 4 contained only 13.2% African-American and Hispanic population. See id.
11. Ellis County, Texas: While the part of Ellis County that was split into Democrat Martin Frost’s District 24 contained 29.0% African-American and Hispanic population, the part split into District 6 contained only 10.8% African-American and Hispanic population. See id.
12. McCulloch County, Texas: While the part of McCulloch County that was split into Democrat Chet Edwards’ District 11 contained 30.3% African-American and Hispanic population, the part of the county split into Republican Lamar Smith’s District 21 contained only 13.1% African-American and Hispanic population. See id.
13. Williamson County, Texas: While the part of Williamson County that was split into District 14 contained 33.7% African-American and Hispanic population, the part split into Congressional District 21 contained only 15.5% African-American and Hispanic population. See id.
14. Tom Green County, Texas: While the part of Tom Green County split into Democrat Charles Stenholm’s District 17 contained 44.6% African-American and Hispanic population, the part split into District 21 contained only 13.1% African-American and Hispanic population. See id.
15. Fort Bend County, Texas: The part of Fort Bend County that was split into Democrat Mike Andrews’ District 25 contained 80.2% African-American and Hispanic population, while the part of Fort Bend County split into District 22 contained only 29% African-American and Hispanic population. See id.
Another split of a different sort merits mention, specifically the splitting of the city of Amarillo, which lies in both Potter and Randall Counties. Randall County and its portion of Amarillo were included in Republican Larry Combest’s District 19, while Potter County and its share of the city were sent to Democrat Bill Sarpalius’ District 13. During debate on the Senate floor on the day of the plan’s passage, Senator Johnson explained the split of Amarillo as a response to the demands of the minority communities in the city:
SIBLEY: You mentioned the Black population of ah Amarillo was moved into the 13th Congressional District. For what purpose?
JOHNSON: Their direct quote to me, and [sic] was that they would rather be in Potter and have a Representative that they could relate to best and the—the respect to the other one in Randall, and they felt that with the two the various populations could communicate well and they had two good team members to look after their interests.
United States Exh. 1092 at 22; see also Dep. of Johnson at 243.
D. Expert Testimony
This portion of the opinion provides a brief summary of the conclusions of most of the experts employed by the parties. Repeating or summarizing the methodology employed by the experts is avoided in the interest of brevity. References to the voluminous reports submitted by the experts will point the interested reader in the right direction.
Dr. Ronald Weber was the plaintiffs’ main expert witness and their only expert to testify at trial. Dr. Weber made the following conclusions in his report: race was the overriding factor in the drawing of lines for Texas Congressional districts; Plan C657 violates traditional districting criteria by creating noncompaet districts as well as by splitting. *132935 counties and a number of municipalities; districts 15, 18, 20, 28, and 30 are “overly safe” from the standpoint of assuring the election of a candidate of choice of African-American or Hispanic voters. See Plaintiff Exh. 36. Finally, Dr. Weber concluded with an analysis of alternative districting plans, including Plan C676 which he developed for this litigation.35 See id. at 30.
Dr. Weber’s conclusions as to the compactness of the Congressional districts merit elaboration. Employing an “eyeball” approach to compactness, Dr. Weber concludes that Districts 3, 4, 5, 6, 8, 9,12,14,15,16,18, 19, 21, 22, 23, 24, 25, 28, 29, and 30 are not compact. See id. at 10. Next, employing three quantitative measures of compactness, dispersion compactness, perimeter compactness, and population compactness, Dr. Weber observes that ten of the districts—3, 4, 6,14, 18, 21, 23, 25, 29, and 30—score in the lowest category on one of the three compactness measures. See id. at 11; see also Pildes & Niemi, supra at 549-50, 553-59 (1993) (assessing worth of each quantitative measure in Shaw-type inquiry).
Dr. Allan J. Lichtman served as the main expert for the State of Texas. His report offers three major conclusions. First, Dr. Lichtman concludes that Plan C657 “substantially protects incumbents of both the Democratic and Republican parties.” State Exh. 14 at 6. Second, elections in the areas in which the state created majority-minority districts are characterized by racially polarized voting between Anglos and minorities. See id. at 4. Finally, Dr. Lichtman suggests that the three alternative plans proposed by plaintiffs—Dr. Weber’s Plan C676, Senator Johnson’s original Plan C500, and the Owens-Pate Plan C606—are deficient in that they “fall well short of the legislative goal of protecting incumbents, both Democrats and Republicans,” id. at 5, and effectively reduce opportunities for minorities to elect candidates of choice. See Plaintiff Exh. 35 (Plan C676); Plaintiff Exh. 33 (Owens-Pate Plan C606); Plaintiff Exh. 29 (Plan C500).
Two other experts for the state submitted reports. Dr. Chandler Davidson discussed the history of African-American voting rights in Texas. See State Exh. 17. He concluded that racial polarization “in the sociological sense” still exists in Texas and that such polarization will actually increase if the number of majority-minority districts decreases. See id. at 87. Dr. Paul Geisel provided a detailed demographic profile of Districts 18, 29, and 30. See State Exh. 18.
The Lawson defendant-intervenors submitted the report of Dr. Richard Murray. See Lawson Exh. 26. Dr. Murray reviewed the history of Congressional redistricting in Harris County, focusing especially on the 1991 redistricting of the county. See id. at 1. Dr. Murray observed that various factors influenced the Legislature in designing Districts 18 and 29: a clear commitment to improve the representational opportunities for Hispanics; the personal ambitions of certain members of the Harris County delegation; protection of incumbents; party politics; class interests; preservation of the 18th as an African-American majority seat; and keeping certain neighborhoods together. See id. at 17-18.
The LULAC defendant-intervenors submitted the report of Dr. Robert Brischetto. See LULAC Exh. 19. Dr. Brischetto’s report evaluates Dr. Weber’s alternative plan and concludes that it “discriminates against Hispanic voters.” See id. at 16. Dr. Brisehetto suggests that the alternative plan attempts to “crack,” “pack,” and “stack” the minority populations in South Texas. See id.
Finally, the United States submitted the reports of four experts. Dr. Paul Waddell conducted an analysis of land uses as related to the boundaries of Congressional District 30. See United States Exh. 1070. Dr. J. Morgan Kousser examined the motivations of recent Congressional redistricting efforts in Texas. See United States Exh. 1071. In reviewing 1991 redistricting efforts, Dr. Kousser essentially concludes:36
*1330It is conceivable that there are instances in which a desire to create a district for one racial or ethnic group is the only explanation for the shape of legislative boundaries. The evidence shows conclusively that Texas in 1991 is not such an instance.
Id. at 53. Dr. Lisa Handley evaluated the districts under Plan C657 in terms of their geographic compactness and concluded that the plan is “reasonably compact because a large portion of the districts are composed of whole counties and follow state, county, and city boundaries.”37 United States Exh. 1067 at 5. Lastly, Dr. Charles Cotrell summarizes § 5 objections in Texas since the state became a covered jurisdiction and considers this historical backdrop in light of the present district configuration. See United States Exh. 1065.
E. Other Districting Plans
Three alternative districting plans received considerable attention in this litigation and will therefore be briefly described. The first was authored by Dr. Weber, the plaintiffs’ main expert witness, as part of his report. See Plaintiff Exh. 35. According to Dr. Weber, Plan C676 allows two African-American districts—one each in Dallas County and Harris County; six Hispanic districts—one in far South Texas, one in El Paso County, two in Bexar County, one in South Texas with Nueces County as its “major center,” and one in rural southwest Texas; and one district in Harris County “subject to substantial African-American and Hispanic influence.” Plaintiff Exh. 36 at 31. Each of these majority-minority districts “[is] also designed to be more compact than the comparable districts in the current state plan.” Id. at 32.
Dr. Weber described the principles he followed in developing Plan C767:
I tried to follow the principle that the metropolitan counties would be split as little as they could, but recognizing, of course, you have to construct whole districts in each of those. What I tried to do is do as few splits between counties, and I got down to basically the point where you have to do some because you have one person, one vote considerations; but basically one of the most important things I did is I told the operator of the computer, I said: “I do not want you to put on the screen the racial make-up of any of the building blocks we are going to use.”
6/28/94 TR. at 133. Although he did not want to split any VTD’s, Dr. Weber testified that he eventually had to for one-person, one-vote considerations. See id.
William Owens and A.J. Pate, two concerned citizens of Texas, presented their plan—eventually denominated C606—to the Texas House Redistricting Committee at hearings in Austin. See Plaintiff Exh. 33. Their plan—entitled “A Modest Proposal for Fair Redistricting in Texas for the 1990’s”— creates a new urban African-American seat in Dallas, “with the possibility of creating a new rural Black impact district.” Id. at 11. The Owens-Pate plan also creates two new Hispanic seats, one in the Houston area and an additional seat along the border with a population base in Laredo. See id. at 3. They believe their plan results in “compact districts of basically equal population, which unifies communities of similar economic/geographic interests” as well as corrects for the underrepresentation of ethnic and racial minorities. See id. at 11.
Senator Johnson’s plan—denominated C500—was first released in May 1991 and drew a great deal of reaction from around the state. See Dep. of Johnson at 51. While minority voters did not object, certain incumbents—especially Congressman Martin Frost—were upset. See id. Under C500, Congressman Frost’s residence was not in District 24 but was included in Senator Johnson’s proposed African-American District 30. See id. at 52. Under the plan, five other congressmen would have been thrown into *1331districts other than the ones they currently represent. See United States Exh. 1036 (“Frost, Bryant Aides Assail New House District Plan”, Dallas Morning News, 5/13/91). Dr. Weber was impressed with C500:
I was really amazed that the very first plan that was released to the public was in my mind the plan that the legislature if, all other things being equal, if they had to rush quickly, they could have adopted it and they would have had a fairer plan than the plan the legislature adopted.
6/28/94 TR. at 129.
IV. FACTUAL FINDINGS AND LEGAL CONCLUSIONS
This court has jurisdiction pursuant to 28 U.S.C. §§ 1331, 1343, 2284. Having considered the evidence, the memoranda of law submitted by the parties, the stipulations of fact, the proposed findings and conclusions, and counsel’s oral arguments, this court makes the following findings of fact and conclusions of law, pursuant to Fed.R.Civ.P. 52(a). Any conclusion of law which should be construed as a finding of fact is hereby adopted as such. Any finding of fact that should be construed as a conclusion of law is hereby adopted as such.
Plaintiffs contend that under Shaw v. Reno, — U.S. -, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993) and the Equal Protection Clause, all but six of the State’s Congressional districts are illegally constituted. They allege that Congressional Districts 18, 29, and 30 owed their extraordinarily odd shapes to an intent to segregate minority voters. These districts, together with District 28, will hereinafter be referred to occasionally as the “voting rights districts.” Other districts in the state, according to the plaintiffs, are the products of intentional segregation because they split counties and cities along racial lines to achieve population balance.38
To evaluate these contentions it is necessary first to review the criteria of a Shaw claim and to weigh some of the State’s general defenses. We then analyze separately the voting rights districts and the State’s other Congressional Districts that are challenged by the plaintiffs. The pertinent issues in each instance are what role race played in the formulation of the districts and whether the resulting districts’ boundaries are sufficiently explainable on other than racial grounds. Finally, we consider whether the state had a compelling justification to segregate voters by race.
The way in which the Court described the nature of the equal protection claim both places Shaw squarely in the traditional mode of constitutional analysis concerning racial classifications and reflects the Court’s sensitivity to the legislative districting process. For reasons that we shall explain, we do not agree with the narrow view that Shaw recognizes an equal protection claim only in such extreme circumstances of racial gerrymandering that hardly any such claim will ever be provable. On the contrary, the Court itself distinguished Shaw on an obvious ground from the truly narrow constitutional claim of partisan gerrymandering adopted in Davis v. Bandemer, 478 U.S. 109, 118-27, 106 S.Ct. 2797, 2802-07, 92 L.Ed.2d 85 (1986): racial classifications are accorded the strictest constitutional scrutiny. See Shaw, — U.S. at -, 113 S.Ct. at 2828.
What, then, does Shaw identify as the characteristics of equal protection in the legislative districting process? The Court accepted the claim of registered voters in North Carolina that “redistricting legislation that is so extremely irregular on its face that it rationally can be viewed only as an effort *1332to segregate the races for purposes of voting, without regard for traditional districting principles and without sufficiently compelling justification” is unconstitutional. Id. at -, 113 S.Ct. at 2824.
The central purpose of the Fourteenth Amendment “is to prevent the states from purposefully discriminating between individuals on the basis of race.” Id. (citing Washington v. Davis, 426 U.S. 229, 239, 96 S.Ct. 2040, 2047, 48 L.Ed.2d 597 (1976)). Drawing on traditional Fourteenth Amendment precedent in racial cases, the Court pointed out that benign or remedial racial classifications are as suspect as malign discrimination and that among the vices of racial classifications is their tendency to “stigmatize individuals by reason of their membership in a racial group and to incite racial hostility.” Id. (citing Richmond v. J.A Croson Co., 488 U.S. 469, 493, 109 S.Ct. 706, 721, 102 L.Ed.2d 854 (1989) (plurality opinion)). Staté legislation that- expressly distinguishes among citizens because of their race must be narrowly tailored to further a compelling governmental interest. See Wygant v. Jackson Bd. of Ed., 476 U.S. 267, 277-78, 106 S.Ct. 1842, 1848-49, 90 L.Ed.2d 260 (1986) (plurality opinion). These analytical principles apply not only to legislation that makes explicit racial distinctions but also to those “rare” statutes that, although race-neutral, are on their face “ ‘unexplainable on grounds other than race.’” See Shaw, — U.S. at -, 113 S.Ct. at 2825 (quoting Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 266, 97 S.Ct. 555, 564, 50 L.Ed.2d 450 (1977)); see also Yick Wo v. Hopkins, 118 U.S. 356, 373-74, 6 S.Ct. 1064, 1072-73, 30 L.Ed. 220 (1886).
The Court agreed with the plaintiff voters’ assertion that if redistricting legislation is so bizarre on its face “that [it] is ‘unexplainable on grounds other than race,’ it demands the same close scrutiny that we give other state laws that classify citizens by race.” Shaw, — U.S. at ---, 113 S.Ct. at 2824-25. The Court cited voting rights cases to support that conclusion. Once established, the Court said, a racial gerrymander should not receive less scrutiny under the Equal Protection Clause than other state legislation classifying citizens by race. See id. at -, 113 S.Ct. at 2826.
Having connected the constitutional claim recognized in Shaw to its unbroken line of Fourteenth Amendment jurisprudence, the Court drew on two voting rights cases and a hypothetical example to illustrate how to recognize a racial gerrymander. Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960), represents the “exceptional ease” in which proof of intentional classification is easy. See Shaw, — U.S. at -, 113 U.S. at 2826. In Gomillion, the Court illustrated its opinion with a map of the formerly rectangular shape of Tuskegee, Alabama, as it had been disfigured by an “uncouth 28-sided” municipal boundary line that allegedly fenced out of the city limits all but four or five of the locality’s African-American citizens. Similarly obvious, Shaw observed, “would be a case in which a State concentrated a dispersed minority population in a single district by disregarding traditional districting principles such as compactness, contiguity, and respect for political subdivisions.” Id. at -, 113 S.Ct. at 2827.
Just as Gomillion demonstrates that irregularly shaped districts can be evidence of purpose or intent to segregate voters by race, so in a later case the Supreme Court held that de facto segregation alone is insufficient if other districting criteria contra-diet a discriminatory purpose. See Wright v. Rockefeller, 376 U.S. 52, 84 S.Ct. 603, 11 L.Ed.2d 512 (1964). In Wright, New York state had redistrieted the New York City area to accommodate four seats in Congress rather than the six that had previously existed. As the district lines were drawn, one district was predominantly white, one was predominantly “non-white,” and the others had sizable minority populations. Plaintiffs contended that the minorities had been packed into one district in violation of Gomillion. The Supreme Court, however, accepted the fact findings of the three-judge panel that intent to discriminate racially had not been proven. Wright observed that because of the geographical concentration of nonwhite voters in one area of the county, it would have been difficult to “fix districts so as to have anything like an equal division of *1333these voters among the districts.” Id. 376 U.S. at 57, 84 S.Ct. at 606.
Wright aptly illustrates the limits of a Shaw claim. Both the Supreme Court and three-judge court decisions in Wright make plain that although there were two racially distinct Congressional districts, their borders were not highly irregular; their configuration was logically traceable from the preexisting district boundaries; each district was reasonably compact and contiguous; and indeed, to have split up a minority population that resided primarily in one geographic area would have been difficult to accomplish. See id. at 56-58, 84 S.Ct. at 605-06; Wright v. Rockefeller, 211 F.Supp. 460, 466-68 (S.D.N.Y.1962).
Wright and Gomillion represent the two poles of a potential Shaw claim.39 From Gomillion, it appears that bizarrely shaped districts whose boundaries were created for the purpose of racially segregating voters are unconstitutional. The Court’s hypothetical example likewise condemns districts that bring together a dispersed minority population without regard for traditional districting criteria. If a majority-minority district reasonably adheres to objective districting factors, however, as in Wright, no invidious discrimination exists; that type of district is justified on its own terms apart from the incidental factor of race.40
Texas asserts two general defenses that, if accepted, would undermine essential premises behind Shaw’s definition of unconstitutional race-conscious redistricting. First, the State asserts that its districts cannot be unconstitutionally bizarre in shape because Texas does not have and never has used traditional redistricting principles such as natural geographical boundaries, contiguity, compactness, and conformity to political subdivisions. Second, the State asserts that the districts’ irregular shapes were caused not by racial classification of voters in any instance but by the Congressional delegation’s demands, acceded to by state government, that all incumbent Congressional officeholders be protected.
It is true that Texas has no constitutional or statutory constraints on creating legislative districts. However, the portrait of redistricting history in Texas, as painted by the state, is inaccurate. From the State’s current perspective, successive generations of Texas legislators have eschewed tying districts to pesky constraints like geography, political subdivision boundaries, compactness, and contiguity. Certainly, this state’s vast layout has undoubtedly made it difficult to fit Congressional districts perfectly within single geographic regions. But since 1960, the principle of assigning at least one Congressional seat to each major city has been followed, satisfying obvious geographical and community interests. See Lawson Exh. 26 at *13348, 9, 11 (maps of Harris County districts in 1957, 1965, and 1971).
Texas points to only one Congressional district that was in the past configured in a highly irregular manner. That was Congressional District 6, which spanned an ungainly rural and urban corridor running from Dallas to Houston and was held for many years by Representative “Tiger” Teague. The exception—a mild deviation from traditional districting principles when compared to the 1991 districts—seems to prove the rule that, generally, Texas has not intentionally disregarded traditional districting criteria. A glance at the maps showing the organization of Texas’ Congressional districts in 1980 refutes the State’s argument that it recognizes no state interest in traditional redistricting principles. See Plaintiff Exh. 28A (map of Plan C001).
More fundamentally, the State describes incumbent protection as a “state interest” in redistricting that sufficiently explains otherwise irregular Congressional district boundaries. There is again tension between the state’s contention and the facts of this case. For one thing, no more than two or three incumbent Texas Congressmen were seriously jeopardized by the Legislature creating more minority districts. Additionally, never before have districts been drawn on a block-by-bloek or neighborhood- or town-splitting level to corral voters perceived as sympathetic to incumbents or to exclude opponents of the incumbents. This form of incumbent protection is much different in degree from the generalized, and legitimate, goal of incumbent and seniority protection previously recognized by the Supreme Court. See, e.g., White v. Weiser, 412 U.S. 783, 791, 93 S.Ct. 2348, 2352-53, 37 L.Ed.2d 335 (1973); Gaffney v. Cummings, 412 U.S. 735, 753 n. 18, 93 S.Ct. 2321, 2331 n. 18, 37 L.Ed.2d 298 (1973).
It is important to realize that as enacted in Texas in 1991, many incumbent protection boundaries sabotaged traditional redistricting principles as they routinely divided counties, cities, neighborhoods, and regions.41 For the sake of maintaining or winning seats in the House of Representatives, Congressmen or would-be Congressmen shed hostile groups and potential opponents by fencing them out of their districts. See 6/30/94 TR. at 4-46 (construction of District 18); Plaintiff Exh. 15 at 5 (construction of District 29). The Legislature obligingly carved out districts of apparent supporters of incumbents, as suggested by the incumbents, and then .added appendages to connect their residences to those districts. See III.C.3. supra. The final result seems not one in which the people select their representatives, but in which the representatives have selected the people.42
But in any event, the State’s realization of its goal may not fully undo the traditional principles of districting that Shaw uses as a benchmark. Shaw nowhere refers to incumbent protection as a traditional districting criterion. Shaw acknowledges that compactness, contiguity, respect for political subdivisions, and like criteria—though not constitutionally required—are “objective factors” that may disprove a racial gerrymander claim. See Shaw, — U.S. at -, 113 S.Ct. at 2827. To this extent, Shaw implicitly reaffirms the important interconnection of community and geography and effective representative government in drawing its distinction between those ideal districting criteria and a racial gerrymander that ignores them.43 While these criteria are important *1335in and of themselves, they are critical to Shaw’s calculus; districts that have no logical boundaries except those dictated by race are perceived by voters within and without the districts as existing solely to afford racial representation.44
The talismanic status of incumbent protection in the State’s argument somewhat resembles Alabama’s defense in Gomillion v. Lightfoot, 364 U.S. 339, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960). As Justice Frankfurter described it:
The respondents invoke generalities expressing the state’s unrestricted power— unlimited, that is, by the United States Constitution—to establish, destroy, or reorganize by contraction or expansion its political subdivisions ... [W]e freely recognize the breadth and importance of this aspect of the state’s political power. To exalt this power into an absolute is to misconceive the reach and rule of this court’s decisions____
Id. at 343, 81 S.Ct. at 128. Justice Frankfurter said that the State was divorcing the teaching of other eases from their concrete factual circumstances. He concluded:
[S]uch power [to redistrict], extensive though it is, is met and overcome by the *1336[Fourteenth Amendment] to the Constitution of the United States ... the opposite conclusion, urged upon us by respondents, would sanction the achievement by a state of any impairment of voting rights whatever so long as it was cloaked in the garb of realignment of political subdivisions.
Id. at 845, 81 S.Ct. at 129. Reliance on a Gomillion -like argument ill-befits the State of Texas. Incumbent protection is a valid state interest only to the extent that it is not a pretext for unconstitutional racial gerrymandering.
Finally, notwithstanding the State’s attempts to minimize their significance, racial data were an omnipresent ingredient in the redistricting process. Preparatory to the 1991 legislative session, the REDAPPL system contained the State’s redistricting maps that were capable of displaying every neighborhood in the state down to the street and block level. As soon as they became available following the 1990 census, the racial statistics for each street and block were coordinated with the REDAPPL system so that the programmer could display both kinds of information simultaneously. No other socioeconomic census data were placed on the computer or available to the Legislature.
Chris Sharman, a legislative assistant heavily involved in the drawing of Congressional districts, confirmed the ubiquity of racial data on the REDAPPL computer system. He also testified that other forms of information pertinent to redistricting were available to the Legislature or the Congressional Delegation. Such information included socio-economic inferences drawn from the shape of streets; drive-by knowledge of territory gained by legislative aides; and the intimate familiarity of legislators with their constituents’ neighborhoods. While some political information was available on the computer at the precinct level, none of the partisan voting information was available for block-by-block portions of VTD’s. In other words, the racial information was the most specific information that could be used by those involved in the districting process.
That district lines throughout the state coordinate very closely with racial population boundaries is hardly disputed by the state. The plaintiffs’ expert demonstrated on maps of most of the 34 Texas counties whose boundaries were split among Congressional districts how closely racial and ethnic population data were coordinated with the Congressional boundary lines. In numerous instances, the correlation between race and district boundaries is nearly perfect. In the Dallas Metroplex area and the Harris County area, where three voting rights districts were created, the racial character of the line-drawing is manifest. The borders of Districts 18, 29, and 30 change from block to block, from one side of the street to the other, and traverse streets, bodies of water, and commercially developed areas in seemingly arbitrary fashion until one realizes that those corridors connect minority populations.
More specific details concerning the racial patterns of districting will be developed below. This court felt compelled to reveal the clarity and detail of the racial input into redistricting not simply because of its significance in this case. The question must arise in the future, when census statistics and other political information become even more sophisticated, how far legislators will go in adjusting district lines to protect their incumbency. Just as they micro-manipulated the racial composition of Texas Congressional districts in 1991, they may be enabled by a new body of statistical data to select their voters even more precisely in 2001.
Before proceeding with the discussion of particular districts, it is necessary to state our understanding of the allocation of the burden of proof under Shaw inasmuch as our understanding differs from the advocacy-based approaches of the parties. We agree with the courts in Hays and in Shaw on remand that a Shaw claim should be proved by the method typical of equal protection analysis. Plaintiffs are obliged to present evidence in support of their racial gerrymandering claim as outlined in Shaw. Defendants then have the burden to produce evidence that any districts found to be racially gerrymandered were dictated by a compelling state interest and are narrowly tailored to further that interest. Like the court in Hays, but unlike the majority on remand of Shaw, we believe part of the State’s burden *1337of production required it to demonstrate that the appearance of the racially gerrymandered districts, as well as their existence, were narrowly tailored.45 As in all Constitutional cases, the plaintiffs retain the ultimate burden of proof.
A. The Voting Rights Districts
The defendants concede that Congressional Districts 18, 29, and 30 were created for the purpose of enhancing the opportunity of minority voters to elect minority representatives to Congress. District 18 has been an African-American majority district since 1970; although its African-American population had declined significantly since 1980, there is no evidence that the Legislature sought to redistrict that seat in a racially neutral way. Congressional District 29, also in Houston, is artfully interwoven with Congressional District 18 and responded to the demands of the Hispanic community to create a district for the burgeoning Hispanic population in Harris County.46 District 30 in Dallas was intended to enable Dallas’ large African-American community to elect a Congressman. Finally, District 28 assured that the expanding Hispanic community in South Texas would elect another member of the Texas Congressional delegation.
The demand for “majority-minority” Congressional Districts was repeated by numerous interest groups who appeared before legislative redistricting committee hearings throughout the state in 1990. Then-Senator Eddie Bernice Johnson vigorously asserted the desire of the African-American residents of the Dallas-Fort Worth Metroplex for a African-American Congressional seat there. In Houston, the Híspame banner was carried by a number of groups; prominently involved in the maneuvering was State Representative Roman Martinez, who intended to run for that seat. Then-State Senator Frank Tejeda from San Antonio intended to satisfy South Texas Hispanics’ claims upon a seat by drawing a district to encompass his Congressional aspirations. The record, in short, is replete with proof that the Legislature intended to devise four majority-minority Congressional seats.
This much can hardly be disputed. The State vigorously disagrees, however, with the suggestion that HB1, the districting statute, accomplished a purposeful racial gerrymander. The creation of majority-minority legislative districts, without more, may raise no Constitutional question. See Shaw, — U.S. at -, 113 S.Ct. at 2828 (“We express no view as to whether ‘the intentional creation of majority-minority districts, without more’ always gives rise to an equal protection claim.”). To ascertain whether the act of the state government verged into unconstitutional territory, the court looks to several sources. Among other things, departures from traditional districting principles such as compactness, contiguity, respect for political subdivisions, and communities of interest are important in determining the legislative intent. See Shaw, — U.S. at ---, 113 S.Ct. at 2825-28. Further, the public record created in the process of enacting HB1 and the § 5 preclearance submission by the State are important to determining the legislative intent. See Arlington Heights, 429 U.S. 252, 266-68, 97 S.Ct. 555, 563-65, 50 L.Ed.2d 450 (1977). Our analysis is facilitated by separately discussing Congressional District 30 and Congressional Districts 18 and 29. We will then consider District 28. Finally, we will discuss together the other districts challenged by the plaintiffs.
1. Congressional District 30
In creating District 30, the testimony reveals that the legislators aimed to reach a minority population of at least 50% voting age population, the minimum number thought acceptable to result in the election of an African-American representative. The ultimate shape of the district has been described as a body with tentacles as “complex and attenuated as a series of DNA molecules.” Michael Barone & Grant Ujifusa, The Almanac of American Politics 1277 (1994); see also Appendix (Map of District 30). The district sprawls throughout Dallas County, deliberately excludes the wealthy *1338white neighborhoods of Highland Park and University Park and extends fingers into Collin County, which include the outermost suburbs of Dallas. In Collin County, the district picks up a small African-American neighborhood. The district extends into Tar-rant County only to pick up a small border area with a high African-American concentration. It also reaches out to claim Hamilton Park, an affluent African-American neighborhood surrounded by whites. Part of the district runs along the Trinity River bottom, using it to connect dispersed minority population. Numerous VTD’s were split in order to achieve the population mix required for the district.
There is nothing compact or contiguous about this district. It is at least 25 miles wide and 30 miles long, but those measurements do not begin to reveal the district’s geographic complexity. Lying in the center of one of the heaviest population areas of Texas, the district’s boundaries reveal as much by what they exclude as by the population they include. Congresswoman Johnson was careful to include those African-American neighborhoods in which she felt the' African-American voting participation rate would be high. She desired to exclude other neighborhoods in which, for reasons such as the disability of felony conviction, African-American voting rate participation would be low. See III.C.5. supra. Then-Senator Johnson’s testimony demonstrated that race was the primary consideration in the construction of District 30. She fought with incumbent Democrats Martin Frost and John Bryant to maintain her share of the Dallas African-American vote. The jostling for position was designed to secure a minimum 50% African-American total population.
The extent to which this district differs from a truly compact and contiguous majority-minority district is revealed by the plan that then-Senator Johnson herself submitted to the state Legislature. See Plaintiff Exh. 29. In that plan, a slightly larger portion of the Dallas County African-American population is included within a much more compact geographic area. Identifiable neighborhoods within Dallas remain intact. No VTD’s are split.
The state asserts that the convoluted boundaries of District 30 are explainable on grounds other than race. Using a demographic study compiled specifically for this trial, the State asserts that there are socioeconomic, geographic, and other similarities among the residents of District 30. See United States Exh. 1070. We reject this study for several reasons. First, there is no evidence that the information it contains was available to the Legislature in any organized fashion before District 30 was created. Second, the study is not persuasive as an explanation for the boundaries of District 30. It describes, but it does not harmonize, the information concerning the district, and it does not differentiate the district from surrounding areas. Third, in so far as the study focuses on multi-family living units, it is inconsistent with the map drawers’ expressed desire to include many single family homes.
The Lawson intervenors assert that there is a community of interest among African-American voters which should be regarded as a legitimate criterion for legislative districting. Even if this argument would support the decision of the Legislature to take African-American voters’ group interests into account in the same way that it does the interest of other voters, e.g., suburbanites, city-dwellers, retirees, the argument cannot go too far. Because a large part of this argument is based upon similarities grounded in race, it is particularly vulnerable to the Fourteenth Amendment’s proscription of racial classification in our society. Further, the intervenors’ argument will not support the creation of districts that have no basis in traditional neutral districting criteria. To permit such districts would fly in the face of Shaw.
The State also proffers its policy of incumbent protection as a non-racial reason for the convoluted boundaries of District 30. By all accounts, the infighting among Congressman Frost, Bryant, and then-Senator Johnson for “sympathetic” voters was fierce. As it happens, however, many of the voters being fought over were African-American. The State cannot have it both ways. It cannot say that African-American voters are African-American when they are moved into Dis*1339trict 30, but they are merely “Democrats” when they are deliberately placed in a contiguous district for the purpose of bolstering the re-election chances of other Democrats.47 The court considers it significant that Robert Mansker, an aide to Congressman Martin Frost, who according to witnesses played a key role in developing the Congressional reapportionment plan, vigorously avoided plaintiffs’ subpoena to testify in this ease.48 See Plaintiff Exh. 37. Parenthetically, if incumbent protection was as important to the process as the State’s witnesses testified in deposition and at trial, it is surprising that the State offered virtually no such evidence either in its voluminous § 5 preclearance submission to the Justice Department or in Terrazas v. Slagle, the previous challenge to the 1991 Congressional districts. With regard to District 30, we conclude that the policy of incumbent protection, to the extent it motivated the Legislature, was not a countervailing force against racial gerrymandering. Instead, racial gerrymandering was an essential part of incumbency protection, as African-American voters were deliberately segregated on account of their race among several Congressional districts.
We conclude that the contours of Congressional District 30 are unexplainable in terms other than race. They have no integrity in terms of traditional, neutral redistrieting criteria. Neighborhoods, VTD’s, and individual streets were split to achieve the district’s racial mix. The district was carefully gerrymandered on a racial basis to achieve a certain number of African-American voters; in order to protect incumbents, other African-American voters were deliberately fenced out of District 30 and placed in other districts that are equally “untraditional.” Plaintiffs have carried their burden of proving a racial gerrymander.
2. Congressional Districts 18 and 29
Even more than in the case of District 30, Congressional Districts 18 and 29 were tailored to include designated numbers of minority voters. Achieving these results was no simple task. African-American and Hispanic populations inhabit large portions of Harris County in a checkerboard pattern with each other and with white citizens. Over the last decade, in fact, former African-American majority District 18 had lost members of that minority population and gained over 40% Hispanic population. Nevertheless, the preservation of District 18 as an African-American majority district was assumed.
A large part of the increase in Harris County Hispanic population during the 1980’s was attributable to immigration. The consequences of immigration for voting rights matters are problematic49 The Legis*1340lature, however, decided to attempt to create a majority-Hispanie Congressional District. The shape of District 29 then became subject to several constraints: the dispersion of the Hispanic population around the borders of District 18 and far to the east in Harris County and the conflicting ambitions of Representative Martinez and then-Senator Gene Green to run for Congress in the “Hispanic” district. As Representative Martinez testified, the borders of District 29 became increasingly distended as he and Senator Green fought to place their state constituents within the new district. Finally, to the south, Congressman Andrews desired to maintain as many minority constituents as possible in his Democratic district.
These two districts are so tortuously drawn that an 8/t x 11" map does not begin to show their block-by-block district lines. See Appendix (Map of Districts 18 and 29). The districts literally meander from one side of the street to the other and cross major thoroughfares, like Shepherd Drive, numerous times.
To say that this configuration violates traditional redistricting principles is an understatement. See generally Pildes & Niemi, supra at 563-64, 567, 569 (concluding that Districts 18 and 29 among the most noncom-pact in the nation). The magnitude of the violation becomes clear in light of the testimony of plaintiff Edward Blum, who campaigned for Congress in the new District 18 in 1992. Blum and his wife spent months walking the entire district in order to shake hands with the voters. They had to carry a map to identify the district lines, because so often the borders would move from block to block. See 6/27/94 TR. at 20-21. Plaintiff Kenneth Powers who assisted Blum in his campaign testified that voters were confused and frustrated. They did not know why their district had been arbitrarily changed, and they did not know the candidates running for office. See Dep. of Powers at 43-44. The boundaries had become so complex that the county clerk’s office sent the wrong ballots to certain precincts and erroneously counted those votes within District 18. The total number of precincts in Harris County nearly doubled following the 1990 redistricting as a result of the complex new district lines.50 See Plaintiff Exh. 10. In Districts 18 and 29, 60% of the residents live in split precincts after redistricting, and in District 25, over 40% were so affected. See Plaintiff Exh. 34Q.
The result of this line-drawing appears utterly irrational—unless one factors in the overlap between these district boundaries and the racial makeup of their underlying populations. The goal of separating Hispanic and African-American residents from each other and from the white population for purposes of voting led to the creation of these particular districts.
As in the case of District 30, the state posits that the irregular boundaries of Districts 18 and 29 were caused by the demand for incumbent protection rather than racial considerations. We disagree. The essential goal in creating these districts was to segregate Hispanics from African-Americans and both minorities from whites in order to retain at least 50% total African-American population in District 18 and achieve at least 61% total Hispanic population in District 29. Then-State Representative Roman Martinez clearly testified so in his deposition in this case where he said it was particularly neces*1341sary to split VTD’s in order to capture pockets of Hispanic residents for the new district.51 He reiterated his insistence on having at least 61% Hispanic population in that district. Further, as we concluded in regard to District 80, the goal of incumbent protection was itself realized by the deliberate segregation of voters on the basis of race in the Harris County metropolitan area. Incumbent Democrats were fencing minorities into their districts or into the new majority-minority districts, while those same minorities were effectively being removed from Republican incumbents’ districts.
The defendants also contend that African-Americans and Hispanics in Harris County belong to identifiable communities of interest that may be and were recognized as such for districting purposes. This argument is troubling for the same reason noted in regard to District 30. Although the issue of minority cohesion is relevant to a § 2 vote dilution claim, it is another matter entirely, as we previously stated, for a racial or ethnic group to claim an award of representation based on race or ethnicity apart from traditional districting criteria. Moreover, one must question how citizen and non-citizen Hispanics comprise a community of interest—many obvious sociological issues, such as relative educational attainment, competition for similar jobs, taxpayer versus non-taxpayer status, and even linguistic differences were not plumbed before this court. We conclude that whatever may be proven in other eases, Shaw does not permit districting to be based on race or ethnicity in conditions such as these, which violate traditional districting criteria.
Finally, the defendants assert that because Districts 18 and 29 each include residents of similar socioeconomic background and he fully within Harris County, they are sufficiently compact to pass muster. We disagree. As all parties and their expert witnesses agreed, compactness must be a relative measure for Legislative districts. The Congressional Districts in West Texas are compact even though they span hundreds of square miles because they encompass all of the population lying within a distinct, clearly defined area. In a major urban county, compactness makes little sense if considered in terms of geographic sprawl alone, but it seems far more probative when viewed in terms of a city’s or county’s neighborhoods, geopolitical subdivisions, and business location. Adjusting the sense of compactness to the complexity and population density of the urban landscape demonstrates that Districts 18 and 29 are not compact at all. Their contorted shapes are the antithesis of compactness.
Because Districts 18 and 29 are formed in utter disregard for traditional redistricting criteria and because their shapes are ultimately unexplainable on grounds other than the racial quotas established for those districts, they are the product of unconstitutional racial gerrymandering.
3. Narrow Tailoring to Achieve a Compelling State Interest?
The defendants also contend strenuously that Districts 18, 29, and 30, having been created in part to satisfy the State’s duties under the federal Voting Rights Act, are for that reason justifiable under Shaw. This is a subtle but significant misreading. A Shaw claim for denial of equal protection is stated if the state created bizarrely shaped districts for the purpose of racially segregating voters who are geographically and otherwise dispersed according to traditional districting criteria. But Shaw makes it plain *1342that the states’s intention to comply with the Voting Rights Act when it created such districts will not necessarily save it constitutionally; the Voting Rights Act may not be used to sanction the “racial apartheid” that Shaw and the Fourteenth Amendment condemn.
Part IV of Shaw clearly holds, however, that compliance with the Voting Rights Act might be a compelling state interest that, if narrowly tailored, would withstand the strict scrutiny demanded of racial classifications under the Fourteenth Amendment. Part IV of Shaw was written to refute Justice Souter’s dissenting position that advocated less exacting scrutiny of racial gerrymanders than is applied to other types of discrimination under the Fourteenth Amendment. See Shaw, — U.S. at-, 113 S.Ct. at 2830. The Shaw majority rejected his contention.52 Part IV is peppered with the language of compelling state interest as applied to the state’s need to comply with the Voting Rights Act. The difference between the State’s interpretation of Shaw and what Shaw really says lies in the allocation of the burden of production. If, as the defendants contend, Voting Rights Act concerns are an “explanation” that justify bizarrely drawn racial districts, the plaintiffs must prove as part of their case that the districts did not comply with the Voting Rights Act. Because, as we have concluded, the plaintiffs’ burden of production extends solely to the race-consciousness of the districts combined with the disregard of traditional districting criteria, then the State has the burden of producing evidence of narrowly tailoring to achieve its compelling state interest.
Interestingly, Texas does not seriously argue that Districts 18, 29, and 30 are “narrowly tailored” to fulfill the State’s obligations under the Voting Rights Act and would thus withstand the strict scrutiny test. Based on the evidence, this would have been nigh impossible. The State admits that more traditional districts could have been fashioned. At least two proposed redistricting plans for Dallas—Senator Johnson’s Plan C500 and Owens-Pate Plan 606—and two for Houston—Owens-Pate Plan 606 and Dr. Webber’s Plan 676—included far more compact, contiguous majority-minority districts. Defendants contend, however, that these districts probably would have sacrificed one or two incumbent Congressmen, but they cannot and do not contend that preserving incumbents rests on the same compelling interest footing as compliance with the Voting Rights Act. Many witnesses acknowledged that majority-minority districts could have been created in Harris and Dallas counties with more respect for compactness, contiguity, geography, and neighborhood preservation. See Dep. of Johnson at 130-32, 142; 6/28/94 TR. at 2-128-129. Under these circumstances, the State has not carried its burden of production on the issue of narrow tailoring.
In its post-trial brief, the United States adopts a different position, contending not only that the State had a compelling interest in complying with the Voting Rights Act but also that Districts 18, 29, and 30 are narrowly tailored to further that interest.53 It is not obvious to this court that the State54 justifiably feared potential liability under § 2 *1343or § 5 of the Voting Rights Act if it failed to protect District 18 and set aside three new districts—Districts 28, 29, and 30—for minority Congressmen. Nevertheless, for this and other reasons, the Legislature created the districts. According to Shaw, this is permissible if the districts are narrowly tailored to comply with Voting Rights Act concerns. But Shaw cautions: “A reapportionment plan would not be narrowly tailored to the goal of avoiding retrogression if the State went beyond what was reasonably necessary to avoid retrogression.” Shaw, — U.S. at -, 113 S.Ct. at 2831. This caution, we assume, would also apply to a § 2 prophylactic measure.
The United States’ characterization of Districts 18, 29, and 30 as narrowly tailored runs afoul of this caution. In the government’s view, Texas could draw these districts in just about any bizarre shape as long as it attributed their shapes to incumbent protection or another “non-racial” consideration. The United States is also implicitly equating incumbent protection with a compelling state interest, an utterly unjustifiable argument.
Because a Shaw claim embraces the district’s appearance as well as its racial construction, narrow tailoring must take both these elements into account. That is, to be narrowly tailored, a district must have the least possible amount of irregularity in shape, making allowances for traditional districting criteria.55 The United States, by deferring heavily to the state’s choice of *1344boundaries, commits narrow tailoring to an insignificant role and ignores the dispositive fact that alternative plans for Districts 18, 29, and 30 were all much more geographically and otherwise logical than the Swiss cheese plans chosen by the state. Where obvious alternatives to a racially offensive districting scheme exist, the bizarre districts are not narrowly tailored.56
From the foregoing discussion, we conclude that Districts 18, 29, and 30 are the product of unconstitutional gerrymandering and that they are not narrowly tailored to further the State’s concern about compliance wdth the Voting Rights Act.
4. Congressional District 28
There is markedly less evidence in the record concerning the creation of Congressional District 28, a designedly Hispanic district located in the South Texas area. South Texas experienced a dramatic growth in population, largely of Hispanic origin, during the 1980’s, and it was foreseeable that a new Congressional district would be located there. Based on the majority-Hispanic population throughout much of South Texas, it is also not surprising that the new district would have an Hispanic majority. Then-Senator Tejeda influenced the drawing of district lines so that as much of his Bexar County constituency as possible would fall within the district, but its progression south from Bexar County is similar to the configuration of the other Congressional districts in South Texas. When compared with the other districts in Texas, Congressional District 28 is not highly irregularly shaped. Its fingers do jut into the small cities of Seguin and New Braunfels. In so doing, according to the plaintiffs’ maps indicating county racial composition, the district excises the minority, largely Hispanic populations from those cities. The number of voters affected by these extensions is small compared to the size of the district. The Legislature took no extraordinary measures that render this district so out of line with traditional districting criteria as to raise a serious question about racial gerrymandering.
B. Other Congressional Districts
Plaintiffs attack most of the other Congressional districts in the State of Texas *1345as having produced deliberate racial segregation of voters to subserve the goal of incumbent protection. As we noted before, there is an extremely high correlation between the irregular features of many of these districts and the racial populations they are drawn to include or exclude. See III.C.8. supra. At trial, Chris Sharman generally explained away these divisions by saying that the incumbent Congressmen were seeking to corral “Democrats” into their districts. More candid testimony was provided by Congresswoman Eddie Bernice Johnson during the Terrazas case, which preceded the issuance of the Supreme Court’s decision in Shaw v. Reno. In Terrazas, then-Senator Johnson stated that African-American voters from many of these towns had requested to be placed within districts of Democrat Congressmen. We conclude that both race and politics influenced the divisions of these towns between Congressional districts.
It does not follow, however that racial gerrymandering occurred. First, with few exceptions, the outlines of the non-Voting Rights Congressional districts within the state (except those which are contiguous to the districts we have already found unconstitutional), are fairly regular or at least not highly irregular apart from the small racially distinct appendages. Second, in deciding whether voters have been segregated by race, the frame of reference must be considered. It is true that voters within individual cities or counties were often separated along racial lines into the districts of incumbent Republican or Democrat Congressmen. From the standpoint of those districts, however, the addition or subtraction of these minority populations was not proportionately significant; they also gave the Congressmen a toe-hold in such cities and effectively doubled the cities’ representation in Congress. Further, we cannot say that from the perspective of the districts, there was unconstitutional racial gerrymandering.
V. CONCLUSION
Shaw explained the nature of a racial gerrymandering claim under the Fourteenth Amendment. Utilizing Shaw’s precepts, the court has carefully analyzed all of the voluminous evidence produced by the parties and investigated 24 of Texas’ Congressional Districts. Although the State indisputably used racial data in the process of Congressional reapportionment throughout the state, and it used the data with sophistication and precision, we conclude that only three Congressional Districts were unconstitutionally racially gerrymandered. Districts 18, 29, and 30 were all designed with highly irregular boundaries that take no heed of traditional districting criteria; those districts function primarily to include sufficient numbers of the favored minority groups and to exclude the disfavored groups so as to assure election of one of the favored groups’ members. If these districts—tortuously constructed block-by-block and from one side of the street to another across entire counties to satisfy the desired racial goal—are constitutional, then the State could more easily hand each voter a racial identification card and allow him to participate in racially separate elections. The exclusively racial makeup of these districts harks back to the infamous “white primary,” which was constitutionally condemned decades ago. See Terry v. Adams, 345 U.S. 461, 73 S.Ct. 809, 97 L.Ed. 1152 (1953); Nixon v. Herndon, 273 U.S. 536, 47 S.Ct. 446, 71 L.Ed. 759 (1927). Surely districts as race-specific as Districts 18, 29, and 30 have no place .in our system of government.
Moreover, Districts 18, 29, and 30 were not narrowly tailored to fulfill the State’s compelling interest in avoiding liability under § 2 or § 5 of the federal Voting Rights Act.
Based on the foregoing the court hereby
ORDERS that Districts 18, 29, and 30 as enacted in HB1 in 1991 are declared unconstitutional under the Fourteenth Amendment. Further relief, consistent with this opinion, will be considered upon written submission by the parties on or before August 26, 1994.
. HBl will alternatively be referred to as Plan C657, the plan number assigned by the State’s redistricting software to the plan embodied in HBl.
. As Texas is covered by § 5 of the Voting Rights Act, the Legislature must either (1) have any proposed plan precleared by the Department of Justice, or (2) seek a judgment from the United States District Court for the District of Columbia declaring that the plan "does not have the purpose and will not have the effect of denying or abridging the right to vote on account of race or color....” Voting Rights Act of 1965, 42 U.S.C. § 1973c.
. The plaintiffs' post-trial submissions seem to suggest that they now challenge all 30 districts, but we reject this belated attempt to broaden the scope of the case.
. To call these districts "configured” in any sense that implies order would be a misnomer.
. In addition to their claim under the Equal Protection Clause of the Fourteenth Amendment, the plaintiffs alleged that the 1991 Congressional Redistricting Plan violated the Fifth and Fifteenth Amendments, as well as the Voting Rights Act of 1965, as amended, 42 U.S.C. § 1973. Complaint at 14 ¶ 39.
. Voting Rights Act of 1965, 42 U.S.C. § 1973.
. As delineated in their June 16th specification, the plaintiffs challenge twenty districts as unconstitutional racial gerrymanders under the framework of Shaw v. Reno, - U.S. -, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). The targeted districts are Districts 3-9, 12-13, 18-19, 21-26, and 28-30. Four additional districts—Districts 1, 2, 14, and 15—were challenged under Hays v. Louisiana (Hays I), 839 F.Supp. 1188 (W.D.La.1993). The Supreme Court vacated the judgment in Hays I and remanded the case to the district court for further consideration in light of the Louisiana Legislature’s repeal of Act 42 and creation of a new districting scheme in Act 1. See Louisiana v. Hays, - U.S. -, 114 S.Ct. 2731, 129 L.Ed.2d 853 (1994). After a two-day trial, the district court once again struck down the Louisiana redistricting plan as an unconstitutional racial gerrymander. See Hays v. Louisiana (Hays II), 862 F.Supp. 119 (W.D.La.1994). The court adopted by reference its opinion- in Hays I. See id. at 121. This court finds it unnecessary to determine whether, as the parties argue, Hays I goes beyond Shaw.
.The parties, however, chose not to use all of their allotted tried time.
. Ted Lyon, a former member of the Texas House and Senate involved in the 1980 and 1990 redistrieting battles, asserted that ”[C]ompactness is not a ‘traditional districting principle’ in Texas. For the most part, the only traditional districting principles that have ever operated here are that incumbents are protected and each party grabs as much as it can. There is no reason why the State should now have to draw compact majority-minority districts when it has shown no interest over the years in drawing compact majority-white districts.” Lawson Exh. 14, ¶ 17; see id. at ¶ 12 ("Neither pretty districts nor compact districts are a priority in Texas, and they have not been since well before I was involved in districting, if ever.").
At a minimum, however, a comparison of Plan C657 and Plan C001, the 1980's Congressional districting plan, strongly suggests that compactness as measured by an “eyeball” approach was much less important in Plan C657. This is especially true of the major urban counties, namely Dallas and Harris. Cf. Plaintiff Exh. 28A (map of Plan C001) with Plaintiff Exh. 34B (map of Plan C657).
. The staff for the Senate Subcommittee on Legislative Redistrieting included Carl Reynolds, Chris Sharman, Shannon Noble, and Laura McElroy. See Stip. 27. Sharman was the subcommittee’s technician and map drawer. See 6/29/94 TR. at 3-156-58.
. In reference to a statement made in the "gray books" that “the Gingles standard does not appear to require a majority district to be drawn if the district would be extremely elongated or otherwise bizarre in shape," Jeff Archer testified that "those terms are to some extent attempts to explain to a person who’s not familiar with this [redistrieting] what compactness might mean.” He noted that the statement was made in the context of a plaintiff in a redistrieting case under *1315section 2. If the best the voters alleging the violation could do was to show that the legislators could have connected South Texas with Houston, and then gone over to Dallas, in his view, the State is probably not under any obligation and is not violating the law by drawing some other forms of districts. See Dep. of Archer at 192-193.
. The agreement on Dallas County traversed party lines. Republican House member Kent Grusendorf testified that fairness and the Voting Rights Act dictated that African-American voters should have a district in Dallas in which they could elect a candidate of choice. See Dep. of Grusendorf at 47.
. Creation of a "safe” African-American district in Dallas County obviously impacted the other districts in Dallas and Tarrant Counties. The Narrative briefly discusses why the splitting of the African-American community in Tarrant County between Districts 12 and 24 does not amount to dilution of that minority community. See id. at 3.
. Harris County Senator Rodney Ellis described the value of majority-minority districts such as District 18:
Majorityf-minority] districts are important to provide opportunities for minorities to elect candidates of their choice. Democracy cannot function at its best when whole groups are excluded or separated from the political process. Majority-minority districts offer people whose philosophies have not always been represented a chance to get into the game. The districts do not ensure that minorities will win seats. They ensure that their viewpoints will be represented. Often these viewpoints are better represented by a minority, but' not always.
Lawson Exh. 7, ¶4.
. The testimony of Harris County Senator Rodney Ellis is consistent with the view that in Harris County African-Americans and Hispanics may not form political coalitions. See Lawson Exh. 7, ¶ 19 (“[T]he candidate of choice of the African-American community will often not be the candidate of choice of the Hispanic community.”)
. For our purposes, a VTD is the functional equivalent of a voting precinct.
. The REDAPPL software did not itself contain electoral databases, but election information was available through the State’s mainframe computer.
.Election information at the block level was available—if at all—only through the personal knowledge of Congressional incumbents and their staff. 6/29/94 TR. at 177-78.
. Newspaper articles appearing statewide before and during the redistricting process confirm the view that the Legislature had early on agreed that Dallas County would get a new minority— namely African-American—district. See, e.g., United States Exh. 1012 (“Redistricting Mostly Pluses for Democrats”, Austin American-Statesman, 4/21/91); United States Exh. 1017 ("Minority District Foreseen for County”, Dallas Morning News, 12/27/90); United States Exh. 1038 ("Minority Seat May Spell Peril for 3 Congressmen”, Dallas Times Herald, 5/5/91).
. It was widely acknowledged that then-Senator Johnson had enormous authority in drawing the boundaries of District 30 as she saw fit:
[A]t some point the lieutenant governor made it clear that he wanted Senator Johnson to draw her district and that the Senate, as far as the lieutenant governor was concerned, was going to support what she wanted to do in Dallas. And at that point it came down to drawing her district, or the district that she would run in, and working with accommodating Bryant on the east side and Frost more or less on the west side to—to accomplish what they felt that they needed to do.
Dep. of Reynolds at 23.
. Congressmen Frost and Bryant had a significant part of their old districts—24 and 5 respectively—removed to draw the safe African-American seat. As Chris Sharman noted: "[A] good portion of the core of District 30 was in Congressman Bryant's district prior, and the other portion was in Congressman Frost’s prior to redistricting....” June 29, 1994 TR. at 3-187.
. Congresswoman Johnson in fact drew a much more compact African-American majority district in Dallas in her Plan C500. See Plaintiff Exh. 29. That plan drew much opposition from incumbents and was quickly abandoned. See Part III. E. infra.
. Ted Lyon was certainly not the sole source of additional support for the proposition that incumbency protection played a role in the boundaries of District 30. Representative Grusendorf testified that the odd configuration of District 30 was the result of protecting Frost and Bryant. Dep. of Grusendorf at 41. And Carl Reynolds, a staff member of the Senate Subcommittee on Legislative Redistricting, testified that there was a conflict over the Dallas area because
Frost and Bryant split Dallas County and we were forcing another district right down in-between the two of them and pushing them outward, and so then naturally there were neighborhoods that one or the other had represented and there were people that—you know, there were people that they had represented and that they wanted to continue to represent and they had to work out the lines to do that.
Dep. of Reynolds at 24-25.
. That District 18—a safe African-American district—ended up having an Hispanic plurality in total population is not surprising given the explosive growth of the Hispanic community in Harris County over the 1980's. See part II.A. supra.
. The maps in the appendix to the opinion roughly outline the districts' severely convoluted boundaries. See Appendix (Maps of Districts 18 and 29).
.The chairman of the Texas Democratic Party, Bob Slagle, told the House Redistricting Committee that an Hispanic Congressional district could be drawn in Harris County while keeping District 18 an African-American district and leaving District 25 with a "significant” minority population of about 40%. See United States Exh. 1058 ("Hispanic District Feasible, House Committee Hears”, Houston Post, 4/2/91). The chairman of the Texas Republican Party, Fred Meyer, expressed his support for efforts to draw the Hispanic district in Harris County, calling plans for such a district “ ‘possible, feasible, responsible and fair.' ” United States Exh. 1061 (“Hispanic Congressional District for Harris County Proposed”, Dallas Morning News, 3/24/91).
. Both Senator Green and Representative Martinez had Congressional aspirations and, consequently, wanted the new Hispanic district to be based around their legislative districts. See Dep. of Martinez at 13-14, 16-18. Green eventually won the race for the District 29 seat in a bitter contest against longtime Houston City Councilman Ben Reyes.
. Senator Rodney Ellis, Democrat from Houston, observed that to a great extent "District 18 was shaped by the districts around it," Lawson Exh. 7 at ¶ 17, in part because it was assumed that District 18 would continue to be an African-American majority district and because incum*1325bent Congressman Craig Washington was not active in redistricting. See id.
. Senator Ellis similarly testified that "[i]n the absence of an Andrews’ district, Senate District [13] could have been the starting point for District 18, with population added to bring it up to the size of a Congressional district.” Lawson Exh. 7 at ¶ 18.
. At least one observer, Ed Martin, noted that, in drawing Districts 18 and 29, "the only way to protect all incumbents of both parties, draw two majority-minority districts, and get 76 votes in the House and 21 votes in the Senate was for one district to go through Downtown and the other to go around it.” Lawson Exh. 15, ¶ 18.
. As Carl Reynolds testified, once a precinct or VTD is split, the redistricting software provides no election information at the block level. See Dep. of Reynolds at 34. Unlike election data, racial information is available at this level. See id. at 35.
. In such micro-precincts, a voter might perceive that the secrecy of his or her ballot was jeopardized by the new precinct lines, especially if turnout was 50% or below.
. A 1982 objection by the Justice Department to proposed east-west configurations for Districts 15 and 27 in South Texas resulted in their present north-south configuration. See United States *1326Exh. 1065 at 9; 6/29/94 TR. at 3-169. This certainly influenced—perhaps dictated-—the north-south configuration of District 28.
. While the part of Guadalupe County split into District 28 contained a combined 52.3% African-American and Hispanic population, the portion slit into District 21—represented by Republican Congressman Lamar Smith—contained only 23.4% African-American and Hispanic population. See Plaintiff Exh. 36H at 4. Similarly, the part of Comal County split into Congressional District 28 contained 62.1% African-American and Hispanic population, but the portion put in Smith’s district was only 10.6% African-American and Hispanic population. See id.
. An overview of the alternative districting proposals is provided in part E. supra.
. As will be fully set out in part IV of the opinion infra, the court strongly disagrees with Dr. Kousser’s view of the "evidence" before him. Furthermore, Dr. Kousser's inquiry was limited to whether "other motives at least partially explain why the lines ... were drawn where they *1330were" or whether race was literally the only reason for district shapes. See United States Exh. 1071 at 1. In part IV, we elaborate on our disagreement with this implicitly narrow view of a Shaw claim.
. At least with regard to Districts 18, 29, and 30, the suggestion that these districts are “compact” under any reasonable definition of the term is a proposition we flatly reject. See IV. A. 1.-2. infra.
. Both the state and the Lawson Intervenors have challenged the standing of the plaintiffs to question the constitutionality of Congressional districts other than those in which they reside. Their argument seems clearly refuted hy Shaw, in which the plaintiffs, five residents of Durham County, North Carolina, challenged the makeup of the two African-American majority Congressional districts in the state. The plaintiffs resided in one of those districts and in an adjoining, non-African American majority district. By deciding that appellants stated a claim for constitutional relief, the Court inferentially decided they had constitutional standing to sue. See Shaw, - U.S. at -, 113 S.Ct. at 2821. On remand, the three-judge court in Shaw v. Hunt, 861 F.Supp. 408 (E.D.N.C.1994), also concluded that the plaintiffs had standing. We agree with that court’s reasoning and conclusion.
. In thus describing a Shaw claim, we do not mean to suggest that other racially motivated districting or election procedures are no longer constitutionally protected. As Shaw stated, the intentional dilution of minority voters remains unconstitutional under the Fourteenth Amendment. See White v. Regester, 412 U.S. 755, 765-66, 93 S.Ct. 2332, 2339-40, 37 L.Ed.2d 314 (1973). The Fifteenth Amendment proscribes outright denial or abridgement of the right to vote.
. With due respect to the two-judge majority on remand of Shaw, 861 F.Supp. 408 (E.D.N.C. 1994), we disagree strongly that the misshapen boundaries of a racially constructed district are merely prima facie evidence of a constitutional violation and not part of its essence. As we pointed out above, Shaw's reliance upon the contrast among Gomillion, the Court’s dispersed minority hypothetical, and Wright refute this reading. Moreover, the Supreme Court’s insistence that appearances do matter, because tortuously constructed districts foster rather than dispel racial consciousness and stereotyping, see Shaw, - U.S. at ---, 113 S.Ct. at 2827-32, expressly relates to traditional districting principles.
The Shaw remand court’s majority had to minimize Shaw's concern with the appearance of racial districts for obvious reasons. By all accounts, North Carolina's majority-minority districts, like those of Texas, are among the most distorted in the nation. If those districts survive constitutional close scrutiny, then Shaw may be a meaningless exercise.
The remand majority in Shaw discounted traditional districting principles both (1) as an element of the equal protection violation and (2) as a standard by which to compare the narrow tailoring of a district found to be racially gerrymandered. We have commented on the error in the first part of the Shaw remand majority’s analysis. The error in the second part of their rationale is discussed infra n. 55.
. Congressional redistricting in 1980 split ten counties. See Plaintiff Exh. 27. Of those ten counties, Harris, Dallas, and Bexar were split because the population of the county exceeded the number of persons required for a single district. Thirty-four Texas counties were split for the current plan. See id. Only Harris, Dallas, Bexar, Tarrant, Travis, and El Paso were split because the population of those counties exceeded the number of persons required for a single district.
. See Plaintiff Exh. 14 at 6 ¶ 14.
.Traditional, objective districting criteria are a concomitant part of truly "representative” single member districting plans. Organized political activity takes place most effectively within neighborhoods and communities; on a larger scale, these organizing units may evolve into media markets and geographic regions. When natural geographic and political boundaries are arbitrarily cut, the influence of local organizations is seriously diminished. After the civic and veterans groups, labor unions, chambers of commerce, religious congregations, and school *1335boards are subdivided among districts, they can no longer importune their Congressman and expect to wield the same degree of influence that they would if all their members were voters in his district. Similarly, local groups are disadvantaged from effectively organizing in an election campaign because their numbers, money, and neighborhoods are split. Another casualty of abandoning traditional districting principles is likely to be voter participation in the electoral process. A citizen will be discouraged from undertaking grass-roots activity if, for instance, she has attempted to distribute leaflets in her congressman’s district only to find that she could not locate its boundaries.
In influencing the Legislature to draw districts that ensured their electoral success, Texas Congressional incumbents apparently foreswore these principles and opted to rely on their name recognition and a small base of zealous supporters to gain re-election. An even more pernicious tendency would seem to follow their incumbent gerrymandering: as the influence of truly local organizations wanes, that of special interests waxes. Incumbents are no longer as likely to be held accountable by vigilant, organized local interests after those interests have been dispersed. The bedrock principle of self-government, the interdependency of representatives and their constituents, is thus undermined by ignoring traditional districting principles.
. Shaw eloquently stated the constitutional offense created by such racial gerrymandering:
Put differently, we believe that reapportionment is one area in which appearances do matter. A reapportionment plan that includes in one district individuals who belong to the same race, but who are otherwise widely separated by geographical and political boundaries, and who may have little in common with one another but the color of their skin, bears an uncomfortable resemblance to political apartheid. It reinforces the perception that members of the same racial group—regardless of their age, education, economic status, or the community in which they live—think alike, share the same political interests, and will prefer the same candidates at the polls. We have rejected such perceptions elsewhere as impermissible racial stereotypes. By perpetuating such notions, a racial gerrymander may exacerbate the very patterns of racial bloc voting that the majority-minority districting is sometimes said to counteract.
The message that such districting sends to elected representatives is equally pernicious. When a district obviously is created solely to effectuate the perceived common interests of one racial group, elected officials are more likely to believe that their primary obligation is to represent only the members of that group, rather than their constituency as a whole. This is altogether antithetical to our system of representative democracy.
* * * * * *
Justice Souter apparently believes that racial gerrymandering is harmless unless it dilutes a racial group’s voting strength. As we have explained, however, reapportionment legislation that cannot be understood as anything other than an effort to classify and separate voters by race injures voters in other ways. It reinforces racial stereotypes and threatens to undermine our system of representative democracy by signaling to elected officials that they represent a particular racial group rather than their constituency as a whole. Justice Souter does not adequately explain why these harms are not cognizable under the Fourteenth Amendment.
* * sis * *
Racial classifications with respect to voting carry particular dangers. Racial gerrymandering, even for remedial purposes, may balkanize us into competing racial factions; it threatens to carry us further from the goal of a political system in which race no longer matters—a goal that the Fourteenth and Fifteenth Amendments embody, and to which the Nation continues to aspire.
Shaw, — U.S. at---, 113 S.Ct. at 2927-32 (citations omitted) (emphasis added).
. See n. 55 infra.
. Ironically, an "Anglo” candidate beat an Hispanic candidate in the election to fill this seat.
.The State asserts that its position is bolstered by the Fifth Circuit's recent en banc decision in League of United Latin American Citizens v. Clements, 999 F.2d 831 (5th Cir.1993) (en banc), cert. denied, — U.S. -, 114 S.Ct. 878, 127 L.Ed.2d 74 (1994). In that § 2 vote dilution case, the court concluded that the plaintiffs had not proven that racially disparate outcomes in state judicial races were caused by white bloc voting against the preferred candidates of African-Americans or Hispanics. Instead, the proof demonstrated that partisan politics rather than racial politics was the determining influence. The state argues that LULAC requires us to find that the legislature's drawing of district boundary lines was also based on partisan politics rather than race. We disagree. The State's argument compares apples and oranges. In LU-LAC, the plaintiffs' burden was to prove whether race motivated white voters throughout the state. Their proof largely consisted of bivariate ecological regression statistics and other inferential techniques. In this case, however, the legislature’s intent was proven directly, through testimony of legislators, the § 5 preclearance submission, and the use of racial data on the REDAPPL system, as well as inferentially from the makeup of the districts themselves.
. At least one newspaper reported that Congressman Frost acknowledged that Mr. Mansker avoided subpoenas "by disappearing for several days in order to avoid helping Republicans with their case.’’ David Flick, GOP Foe Challenges Frost to Release Aides' Vouchers, Dallas Morning News, July 29, 1994, at 23A.
. Under the Constitution as routinely interpreted, representation in the U.S. House of Representatives is apportioned among the states according to the number of persons—not just citizens—in their population. States with high numbers of non-citizens, like Texas, benefit from this situation.
But the consequences of a large non-citizen population for voting rights issues are unclear and have been side-stepped by the Supreme *1340Court. See Johnson v. DeGrandy, - U.S. -, - n. 18, 114 S.Ct. 2647, 2662 n. 18, 129 L.Ed.2d 775 (1994). In Texas, the overwhelming majority of non-citizens are Hispanics. Former Representative Roman Martinez estimated that 40% of Hispanics residing in Harris County are ineligible to vote because they lack citizenship. This uncomfortable fact did not deter Texas Hispanic politicians from demanding more Hispanic Congressional seats—even though § 2 vote dilution claims are ordinarily premised on measures of citizen voting age population.
We decide this case based on the assumption, shared by the parties, that for equal protection purposes, Hispanics are Hispanics, whether citizens or not. We believe, however, that in future cases, at the minimum, Hispanic plaintiffs should have to prove that the citizen and non-citizen Hispanic populations should be regarded as a cohesive ethnic group. See, e.g., Shaw, - U.S. at -, 113 S.Ct. at 2830; Growe v. Emison, - U.S. -, 113 S.Ct. 1075, 1076, 122 L.Ed.2d 388 (1993).
. Harris County created 553 new precincts, bringing its total from 672 to 1225 for the county. See Plaintiff Exh. 10.
. During Dr. Weber’s testimony at trial, the court asked whether he had analyzed the portion of the split VTD's that were excluded from the districts and compared those numbers with the core of the district from which they were excluded. See 7/1/94 TR. at 26-27. Dr. Weber indicated that he had not performed this analysis but indicated that he had provided the court with the necessary numbers and methodology.
The court has analyzed the racial composition of the split VTD’s in Harris County based on numbers provided by the Texas Legislative Council and included in Dr. Weber's report. In examining the number of individuals from VTD’s that were split in creating Districts 18 and 29, an overwhelming majority of those individuals placed in District 29 were Hispanic while an overwhelming majority placed in District 18 were African-American. This analysis supports the inference that VTD’s in Harris County were split for the central purpose of achieving a certain racial composition.
. For this reason, we reject the argument of the United States that “benign” race-conscious districting is subject to intermediate rather than strict scrutiny. The United States founds its position in part on Metro Broadcasting v. Federal Communications Comm’n., 497 U.S. 547, 110 S.Ct. 2997, 111 L.Ed.2d 445 (1990), which applied that standard to federal antidiscrimination measures. We agree, however, with the Shaw remand court that Metro Broadcasting has little to do with this case. See Shaw v. Hunt, 861 F.Supp. 408, 434 n. 22 (E.D.N.C.1994).
. The United States cites Richmond v. J.A. Croson, 488 U.S. 469, 498-500, 109 S.Ct. 706, 724-25, 102 L.Ed.2d 854 (1989), for the proposition, reiterated in Shaw, that a jurisdiction might enact “affirmative action redistricting” if it had a compelling interest in eradicating particular instances of racial inequality. No evidence was presented at trial to support this basis for minority districts, and we will not consider it further.
.As the State concedes, § 5 of the Act, which implements a racial nonretrogression principle in districting, only required the state to preserve extant "minority” Districts; it did not require the creation of new minority Districts 29 and 30. See Beer v. United States, 425 U.S. 130, 141, 96 S.Ct. 1357, 1363-64, 47 L.Ed.2d 629 (1976).
The defendants also assert that if Districts 18, 29, and 30 had not been drawn as majority-minority districts, the State would have been vulnerable to a vote dilution claim under § 2 of the Voting Rights Act. Further, the Attorney *1343General, exercising her responsibility under § 5 of the Act, could have refused to preclear the State's apportionment plan for this reason. Section 2 vote dilution claims are proven by establishing the three "Gingles criteria" and then by showing dilution under the totality of the circumstances surrounding a districting scheme. See Thornburg v. Gingles, 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986). The first Gingles criterion is that the minority population are sufficiently numerous and geographically compact to form a majority in a single-member district. See Growe v. Emison, - U.S. -, -, 113 S.Ct. 1075, 1084, 122 L.Ed.2d 388 (1993). This aspect of Gingles, like Shaw, presupposes legislative districts that have geographic integrity and satisfy traditional districting standards. See, e.g., Johnson v. DeGrandy, — U.S. -, -, 114 S.Ct. 2647, 2655, 129 L.Ed.2d 775 (1994). That test cannot be met under § 2 for Districts 18, 29, and 30. Defendants' perfunctory and wholly unsupported contentions to the contrary are incorrect. Also undermining the assertion of § 2 violation is the recent Supreme Court decision in Johnson v. DeGrandy, which holds that it is not necessary for a political jurisdiction to maximize minority voting strength in order to comply with § 2. See id. at -, 114 S.Ct. at 2661. As the statute itself says, proportional representation is not its goal or mandate.
Finally, Gingles requires proof that the majority usually vote as a bloc to defeat the minority's preferred candidates. In Texas in the 1990's, it is no longer accurate to assume that this condition of Gingles exists in every case. Johnson v. DeGrandy, supra, emphasizes a similar point: “[T]here are communities in which minority citizens are able to form coalitions with voters from other racial and ethnic groups, having no need to be a majority within a single district in order to elect candidates of their choice. Those candidates may not represent perfection to every minority voter, but minority voters are not immune from the obligation to pull, haul and trade to find common political ground.” Id.; see also, LULAC v. Clements, supra, 999 F.2d 831 (5th Cir.1993) (era banc) (concluding minority judicial candidates defeated not because of race but because of partisanship), cert. denied, - U.S. ---, 114 S.Ct. 878, 127 L.Ed.2d 74 (1994). What amounts to a ritualistic invocation of § 2 by the defendants simply proves too much: the possibility that some § 2 claim might have prevailed against the state if HB1 had not contained more minority districts than the base plan cannot justify these noncompact, noncontiguous districts.
. Regarding this aspect of narrow tailoring, we again register disagreement with the two-judge majority that decided Shaw on remand. Those judges conclude that the only factors pertaining to the shape and size of a district that bear on narrow tailoring are constitutional limits, i.e. compliance with the one person/one vote principle and the right of voters not to have their votes diluted. To these limits, however, must be added Shaw's emphasis on the requirement that a racially constructed district must satisfy other neutral districting criteria. The two-judge majority simply ignored this point, as Judge Voorhees, dissenting, observes. Shawv. Hunt, 861 F.Supp. 408, 479 (E.D.N.C.1994) (Voorhees, J., concurring in part, dissenting in part).
The Shaw remand majority apparently concluded that the shape of a "voting rights district” is immaterial as long as the state had a sufficient basis upon which to believe it might be vulnerable to a § 2 vote dilution claim. This conclusion overlooks the Supreme Court’s clear distinction between "what the law permits, and what it requires.” Shaw, - U.S. at -, 113 S.Ct. at 2830. Further, the Shaw remand majority describe as dicta the Supreme Court’s observation that the deliberate creation of majority-minority districts to remedy past discrimination was only constitutionally permissible beyond the framework of the Voting Rights Act if the state employed sound districting principles and if the racial group's residential patterns permitted the *1344creation of districts. Shaw, - U.S. at -, 113 5.Ct. at 2832 (citing United Jewish Organizations of Williamsburgh, Inc. v. Carey, 430 U.S. 144, 167-68, 97 S.Ct. 996, 1010-11, 51 L.Ed.2d 229 (1977) (opinion of White, J.)). We believe these statements bear importantly on the "fit” between the state’s compelling interest in addressing Voting Rights concerns and the form of the districts that must be narrowly tailored to suit them.
Finally, we note that under the reasoning of the Shaw remand majority, a bizarrely shaped district that has no grounding in traditional districting criteria was held “narrowly tailored," although it is inconceivable that the same district could have been authorized under § 2 and the first prong of the Gingies test. See note 54 supra.
Among their reasons for discounting the significance of compactness and contiguity for narrow tailoring, the Shaw remand majority denigrate the importance of those factors and issue an expression of judicial restraint that is, frankly, hard to swallow. We have already defended objective districting criteria. Supra at n. 43. Moreover, judges are routinely deciding nontraditional questions in § 2 vote dilution cases, among which are the possibility to create "compact and contiguous” minority-majority districts. The likelihood of judicial intrusion upon the state’s prerogative of districting is the same in both cases. When a § 2 or Shaw violation is found, of course, judges must defer as far as possible to the legislature’s attempt to solve the problem. See Upham v. Seamon, 456 U.S. 37, 40-41, 102 S.Ct. 1518, 1520-21, 71 L.Ed.2d 725 (1982). Thus, we believe the Shaw remand majority search in vain for policies to support th'eir decision.
. Courts confronted with the question whether a remedial racial classification is narrowly tailored to serve its purpose have considered a number of factors to be important. These factors include: (1) the efficacy of alternative race-neutral measures; (2) the efficacy of alternative, more narrowly-tailored racial classifications; (3) the flexibility and duration of the remedy; and (4) the impact of the remedy on the rights of third parties. See, e.g., United States v. Paradise, 480 U.S. 149, 171, 107 S.Ct. 1053, 1066, 94 L.Ed.2d 203 (1987); Local 28, Sheet Metal Worker’s v. EEOC, 478 U.S. 421, 481, 106 S.Ct. 3019, 3052-53, 92 L.Ed.2d 344 (1986) (Powell, J., concurring); Fullilove v. Klutznick, 448 U.S. 448, 510-11, 100 S.Ct. 2758, 2791, 65 L.Ed.2d 902 (1980) (Powell, J., concurring); see also Ravitch v. City of New York, 1992 WL 196735 at *7 (S.D.N.Y. Aug. 3, 1992) (applying these factors to a racial classification used in redistricting); Hays II.
Because we have concluded that Districts 18, 29, and 30 are not narrowly tailored to reflect, as far as possible, traditional districting criteria, we need not discuss these factors in detail." Our conclusion bears on the first, second, and fourth factors.