Full opinion text
MEMORANDUM WOODLOCK, District Judge. Two sets of plaintiffs in these consolidated actions claim that the City of Boston and the Commonwealth of Massachusetts, in the 1987 reapportionment of their elected deliberative bodies, violated the federal constitutional mandate of electoral equality expressed by the shorthand phrase “one person one vote.” With the cooperation of the parties, the issues were framed to be presented to this three judge court on an expedited schedule and in phases. The threshold issue presented — as Phase IA — is to resolve the validity of the underlying population figures which provide the basis for the reapportionment of districts. The remaining issues in the case necessarily turn upon resolution of the population validity issue. After trial of Phase IA, we have determined that the underlying population figures used by the defendants for apportioning districts withstand the challenges the plaintiffs have mounted. During the trial of Phase IA, we also took evidence on a contingent basis concerning a corollary issue as Phase IB: the compliance of the 1987 reapportionment of the Massachusetts House of Representatives with constitutional requirements. Our evaluation of the 1987 House Redistricting Plan leads us to the conclusions that the state cannot justify the substantial divergences in population among its reapportioned districts and, accordingly, that the 1987 reapportionment of the Massachusetts House of Representatives is unconstitutional. We set forth herein, pursuant to Fed.R. Civ.P. 52, our findings of fact and conclusions of law as to the trial of Phases IA and IB. I Under the Massachusetts Constitution, as amended in 1975, a census of the inhabitants of the Commonwealth of Massachusetts was required to be undertaken in 1975 and once every ten years thereafter. Mass. Const.Amend. Art. Cl. The purpose of the census, according to the Constitution, is to provide “the basis for determining the [representative, senatorial and councillor] districts.” Id. During the first seven months of 1985, all Massachusetts cities and towns undertook a census to measure their populations as of January 1, 1985. On February 11, 1986, the Mayor of Boston reported a Boston population figure of 620,889 to the Secretary of the Commonwealth. On April 16, the Secretary rejected the City’s figure for two reasons: first, because there was a “substantial discrepancy” between that figure and the 1984 Boston population estimate of 570,719 made by the United States Bureau of the Census, and second, because the results of the Secretary’s own field audit found “statistically significant errors” in the City’s figures. The Secretary ordered Boston to resubmit a corrected census count within two weeks. After several months of wrangling between the City and the Secretary, Boston proposed on August 18, 1986, to conduct a new census to begin immediately. This census was never undertaken. On February 9, 1987, the Governor, acting pursuant to Mass.Gen.L. ch. 9, § 7, newly amended, appointed a five member Decennial Census Commission (“DCC”) to study the Boston census. The duties of the DCC are concisely stated in the statute: “The commission shall determine and verify the census for [any] city or town [whose census has been refused by the Secretary because it ‘contains significant errors’] which then shall be accepted by the secretary.” Between February 17 and May 19, 1987, the DCC received written submissions from both Boston and the Secretary, heard testimony from witnesses, received advice from experts and counsel, and produced its final report. In its report, the DCC found serious methodological errors in the Boston census figures. The DCC rejected the Boston census, calculated the January 1, 1985 Boston population to be 601,095, and reallocated population figures for all Boston wards and precincts. DCC Report at 61-62, 78-81. The DCC submitted its report to the Secretary on May 19, 1987, and the Secretary, in turn, forwarded the DCC Report to the General Court on the same day. On June 18, 1987, acting pursuant to redistricting legislation specifically governing it, Chapter 343 of the Acts of 1986, § 1, the City of Boston reallocated its precincts into city electoral districts in accordance with the DCC Report. Meanwhile, special joint committees of the Senate and the House, charged with recommending apportionment of the General Court and the Executive Council, reviewed the recommendations of the DCC, reporting their findings in separate documents respectively on July 8 for the Senate and the Executive Council and on July 9 for the House of Representatives. With modest changes not relevant here, the General Court thereupon reallocated the House, Senate, and Executive Councillor districts in accordance with the figures supplied by the DCC. Immediately after adoption of the new apportionment plan, the Political Task Force plaintiffs began the litigation of Civil Action No. 87-1886 and the Republican plaintiffs began litigation in Civil Action No. 87-1958. The plaintiffs in both actions seek review of aspects of these 1987 reapportionment decisions prior to the 1988 state elections. II At the outset, we confront the question of what standard of review should apply in evaluating the reapportionment efforts of the defendants. Despite over a quarter century of litigation in this area, uncertainty remains concerning the proper standard of review for reapportionment plans. In the most recent edition of his treatise, Professor Tribe notes that this uncertainty was manifest in the initial reapportionment case, Baker v. Carr, 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962), which opened the doors to federal courts for resolution of reapportionment disputes. L. Tribe, American Constitutional Law (2d ed. 1988) 1063. See also, The Supreme Court, 1982 Term, 97 Harv.L.Rev. 135, 139-40 (1983); Casper, Apportionment and the Right tó Vote: Standards of Judicial Scrutiny, 1973 Sup.Ct.Rev. 1. It is tempting to propose an answer to this question by constructing a pigeonhole for reapportionment cases using one of the several articulated standards of review offered in the various forms of equal protection litigation. But such an answer is not easily available. The Supreme Court’s first state legislative reapportionment decision itself variously sounded in both the “rational relation” and “strict scrutiny” standards of Equal Protection Clause analysis. Reynolds v. Sims, 377 U.S. 533, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964), speaks alternatively of the need to defer to “effectuation of a rational state policy,” id. at 579, 84 S.Ct. at 1391 (emphasis supplied), and the proposition that any alleged infringement of the right to vote “must be carefully and meticulously scrutinized,” id. at 562, 84 S.Ct. at 1381 (emphasis supplied). No clear sanction for the use of particular traditional equal protection standards appears in the case law since Reynolds. Nor does this case require us to choose among traditional standards. That is because, as will be developed more fully below, we conclude after careful and meticulous scrutiny that the state’s choice of underlying population figures is justified. And, by contrast, we conclude that the divergences in House districts fail rationally to effectuate identified state policy. Having found that the state meets the highest standard with respect to population validity and fails to meet the lowest standard with respect to House plan proportionality, we need delineate no further the role standard of review plays in this reapportionment litigation. In outlining our reasons for reaching these conclusions, we are instructed by the quarter century of reapportionment litigation, which — while not formally articulating a particular standard of review — has developed a more discriminating approach to the relevant interests by a balancing of the practical factors called into play during the reapportionment process. The Equal Protection Clause requirement in the context of state reapportionment is broadly stated: “that a State make an honest and good faith effort to construct districts ... as nearly of equal population as is practicable.” Id. at 577, 84 S.Ct. at 1390. Federal courts are prepared to tolerate modest deviations from statistically ideal electoral districts. In this connection a ten percent total deviation, computed by adding the individual deviations of the largest and the smallest districts from the ideal district has been developed to mark the upper limit of the de minimis acceptable range for divergence from population equality. See Brown v. Thomson, 462 U.S. 835, 842-43, 103 S.Ct. 2690, 2696, 77 L.Ed.2d 214 (1983). Justification, however, will be required to uphold deviations of more than ten percent because “a State’s policy urged in justification of disparity in district population, however rational, cannot constitutionally be permitted to emasculate the goal of substantial equality.” Mahan v. Howell, 410 U.S. 315, 326, 93 S.Ct. 979, 986, 35 L.Ed.2d 320 (1973). But once the preeminence of practical population equality is recognized and the range of facially tolerable deviations is identified, the evaluative process becomes intensely fact bound. The showing required to justify [more than de minimis ] population deviations is flexible, depending on the size of the deviations, the importance of the State’s interests, the consistency with which the plan as a whole reflects those interests, and the availability of alternatives that might substantially vindicate those interests yet approximate population equality more closely. Karcher v. Daggett, 462 U.S. 725, 741, 103 S.Ct. 2653, 2664, 77 L.Ed.2d 133 (1983). The task, then, is one of case-by-case analysis. Chapman v. Meier, 420 U.S. 1, 15-18, 23-24, 95 S.Ct. 751, 760-61, 764, 42 L.Ed.2d 766 (1975); Mahan v. Howell, 410 U.S. at 329, 93 S.Ct. at 987. There being no necessity of pursuing further the elusive goal of articulating an all-purpose standard of review for reapportionment cases generally, we turn to the task at hand in the instant case. III. The plaintiffs’ challenge to the reapportionment efforts of Massachusetts in 1987 begins at the foundation of the process: the basis upon which the population to be allocated among districts was determined. In evaluating this challenge, we adopt the observation made by then Chief Judge Caffrey that in determining population for reapportionment purposes there is a common sense judicial preference for “the most recent and accurate measure of population.” Latino Political Action Comm., Inc. v. City of Boston, 568 F.Supp. 1012, 1018 (D.Mass.1983). But we recognize as well — as has more recent case law in this District, MacGovern v. Connolly, 637 F.Supp. 111, 114-15 (D.Mass.1986) (per curiam) (3 judge court: Torruella, Cir. J.; Caffrey, C.J.; Zobel, J.) —that this preference is instructed by a certain deference to the mechanism by which the state goes about obtaining such data. Reynolds v. Sims established at the inception of this branch of federal reapportionment law that a state would meet its obligation “for maintaining a reasonably current scheme of legislative representation,” id. at 584, 84 S.Ct. at 1393, by putting in place “a reasonably conceived plan for periodic readjustment of legislative representation.” Id. at 583, 84 S.Ct. at 1393. Consequently, the threshold question here is whether the Massachusetts plan of ascertaining population for its 1987 reapportionment decisions was reasonably conceived to obtain the most recent and accurate available population data. That question turns on an evaluation of the Decennial Census Commission’s work and the alternatives available to the state. -A- The basic charge of the DCC was the determination and verification of the population of Boston and its precincts and wards. After examining the procedures of the Boston census, and hearing from several witnesses, the DCC concluded that the Boston census was flawed and that the population figures generated by that census could not be verified. DCC Report at 29-43. Similarly, the DCC found it impossible to correct the various procedural and methodological errors committed in the City census. Id. at 44. No party to this action contests these conclusions of the DCC. All parties agree that the best means for obtaining an accurate population figure is through a properly conducted census. Of course it is also clear that even a careful census cannot be expected to determine accurately the population of a city as large as Boston. Id. at 27. Still, the numerical counting process a census employs is much more likely to capture population figures accurately than is a statistical estimate, regardless of the complexity and precision of the statistical model. Whatever the actual advantages of a census, however, the DCC had neither the time, the resources, nor the mandate to conduct a de novo census of Boston. Hence, it was obvious that any population figure proposed by the DCC would necessarily be twice removed from the exact population count for which a census strives: it could not be the product of a new census and it would rely necessarily on statistical extrapolations from prior ones. To determine the January 1,1985 population of Boston, the DCC considered alternative methodologies for deriving the figure. Only two approaches have been represented to us as appropriate for population estimates in the absence of a census. The first is a “top down” procedure that tests a completed census figure by subjecting it to statistical tests to determine error rates or confidence levels. Again, all parties apparently agree that the DCC properly rejected the top down approach here because the Secretary’s audit was the only independent measure of the accuracy of the Boston census, and that the census provided an insufficient basis for a proper population determination. The DCC adopted instead an alternative bottom up methodology. The bottom up approach employs an accepted earlier population figure, usually from a census, and then computes changes in population over time by looking either to (a) the net population changes (+ births, — deaths, + net migration) or (b) the net change in habitable housing stock (multiplying number of housing units times occupancy rate times average household size). In order to determine the total population of Boston, the DCC opted for approach (b), net change in habitable housing stock. The DCC adopted the United States Bureau of the Census 1980 population figure for Boston as the base, and then computed the net change in habitable housing. Following this method, the DCC determined that the population of Boston on January 1, 1985, was 601,095. Id. at 59. All parties to this action accept the DCC’s determination of the total population figure for Boston. The next task of the DCC was to apportion the total population figure among the City’s twenty-two wards and the 252 precincts from which the wards are constituted. Although the DCC first sought to apportion population in the same manner used to determine the total population growth since 1980 (i.e., a bottom up approach using housing figures), it concluded that computing housing figures on a precinct by precinct basis created a margin of error that was unacceptably high. That is to say, although computing net housing for the City was possible because of the statistically significant size of the sample, the sample presented by precincts was small and unreliable. Apparently none of the parties disputes this conclusion of the DCC. It is at the DCC’s next step that the consensus breaks down. The DCC decided to use the results of the previously rejected 1985 Boston census for a limited purpose. The Commissioners assumed that the actual growth within and among each of the wards could be computed by: First, determining a ward’s unadjusted growth by subtracting its 1980 federal census figure from its 1985 Boston census figure; Second, multiplying that raw, unadjusted number by .65 (the percentage by which the DCC had determined that the Boston census overcounted: namely, the difference between the 1980 federal census figure and the 601,095 DCC total Boston figure divided by the difference between the 1980 federal census figure and the 620,889 Boston census figure); and Third, adding that adjusted figure to the 1980 ward census figure; Id. at 61-62. The DCC found, however, that it could not apply the same method it used in calculating ward populations for precinct populations because of reduced statistical reliability. As the DCC observed, “[u]nlike the ward counts, which are large enough to demonstrate relative growth propensity, even very slight undercounts and over-counts have a substantial effect on precinct populations.” Id. at 62. It therefore concluded that it would be most appropriate to allocate a ward’s 1980-85 population growth among the precincts within the individual wards by the same percentage of the ward population each of the precincts represented in 1980. Id. The DCC’s apportionment of ward and precinct populations is the crux of the population allocation dispute both sets of plaintiffs raise. We turn directly to that dispute. -B- The core of the problem, according to the plaintiffs, is that immediately after rejecting the Boston census because of its serious methodological flaws, the DCC reversed itself and employed the ward allocations of the Boston census as a critical variable in the determination of the ward populations. “The result,” in the Political Task Force plaintiffs’ words, “was a flawed set of ward and precinct population numbers which incorporates all the errors of the city census, including undercount of minority groups and areas relative to whites and white areas.” Political Task Force Plaintiffs’ Trial Brief at 2. From the plaintiffs’ perspective, the structure of the DCC methodology resulted in three basic statistical and evidentiary flaws: (a) Adopting the Boston census figures for any purpose created an internal inconsistency with the DCC’s own findings of the invalidity of that census. (b) Errors of overcounting in the Boston census were wrongly assumed by the DCC to have occurred uniformly throughout the City and accordingly were improperly distributed on a pro-rata basis among the wards. (c) Within the wards a uniform rate of growth among all precincts was assumed, with a consequent failure to recognize the dynamics of differential intra-ward growth. These three basic methodological flaws are said by the Political Task Force plaintiffs to have resulted in the DCC’s effectively undercounting minorities. The Republican plaintiffs do not focus their objection so narrowly. Their contention is that given these flaws the DCC population figures are wholly unreliable. In order to understand the precise character of the plaintiffs’ contentions we must examine the alternatives they offer. Inevitably, the goal of obtaining and using “the most recent and accurate measure of population,” Latino Political Action v. Boston, 568 F.Supp. at 1018, introduces the need for a comparative judgment in evaluation of population data. It is only by comparing the plan actually used by the state defendants with other available alternatives that the acceptability of the conception of the state reapportionment plan can be understood and evaluated against any standard of review. -1- The gravamen of the Task Force plaintiffs’ complaint is that under the DCC’s faulty methodology, minorities (Blacks, Hispanics, and Asians) were undercounted in the population and hence are underrepresented in the apportionment. Specifically, they contend that the DCC numbers do not show the relative increase of minority population in relationship to the White population. Id. As an alternative methodology to allocate more accurately the Boston population within its wards and precincts, the Political Task Force plaintiffs rely upon an alternative bottom up methodology proposed by Dr. Camayd-Freixas, a founder of the plaintiff Massachusetts Latino Democratic Committee. The Camayd-Freixas procedure first assumes the accuracy of the 1980 federal census and the DCC’s 1985 population calculation. Second, it derives a race specific growth rate from the net increase (or decrease) of births, deaths, and migration for each of the principal population groups of Boston (White, Black, Hispanic, Asian, and “Other”) during this five-year period. It then adjusts the earlier census figure by multiplying the race-specific growth rates against the 1980 population groups in each of the City’s precincts. We find three specific problems with the Camayd-Freixas study. First, the derivation of the race-specific growth rates appears arbitrary. Second, the uniform application of growth rates by race to all precincts does not recognize the different racial growth patterns and migration within the city. Third, the growth rate methodology and figures appear to be systematically self-serving. Despite obstacles to providing hard figures for calculating growth rates generally for a population — as demonstrated by the DCC’s efforts to validate a total Boston population figure and the uneasiness which surrounds its reception, see note 13 supra —Dr. Camayd-Freixas purports to be able to determine an even more difficult and dynamic demographic concept, the relative growth rates for Boston’s racial populations in the period between 1980 and 1985. These racial growth rates Dr. Camayd-Freixas posits as follows: Whites 1.01515 Blacks 1.13966 Hispanics 1.28713 Asians 1.24979 Other 1.18994 Because racial growth rates are the decisive determinants of the Camayd-Freixas allocation of Boston’s population, it is necessary to examine with some care how these figures were derived. We credit the explanation provided by the defendants’ expert, Dr. Ericksen, regarding the manner by which Dr. Camayd-Freixas derived his estimates. Dr. Erick-sen demonstrates that the growth rates were derived by Dr. Camayd-Freixas in part from an arbitrary selection of racial growth rates contained principally in different projections by the Boston Redevelopment Authority (“BRA”). Rather than consistently following the assumptions of a given projection, Dr. Camayd-Freixas selected growth rates from separate studies with differing basic assumptions projecting well past 1985 — but always in a manner that appeared to demonstrate maximization of Black and Hispanic growth and minimization of White population growth. Dr. Camayd-Freixas criticized the DCC study for providing insufficient evidence for its assumptions about similarities in growth rates in wards and precincts. Yet Dr. Camayd-Freixas made methodologically similar assumptions about uniformity of growth rates of racial groups, ignoring any data about differences among growth rates of same-race groups in different precincts and wards, or across economic, educational, or social strata. Although the blanket assumptions about the evenness of growth within wards made by the DCC may be methodologically gross and unsatisfying, the finer calculations made by Dr. Ca-mayd-Freixas based on race are even less acceptable. Those calculations call instead for highly speculative assumptions about racial growth in very small samplings of the population — within precincts. The alternative figures proposed by the Political Task Force plaintiffs rest precariously on the erroneous assumption of identical percentage growth in population by race in each ward and precinct of Boston. The implication of this error is perhaps best seen in Roxbury. BRA statistics, upon which Dr. Camayd-Freixas is prepared to rely when convenient, show a decline of eight percent by Black inhabitants in Rox-bury since 1980. Yet, the Camayd-Freixas methodology, which assumes uniform increases in Black population of 14.6 percent throughout Boston, projects a substantial increase in the Black population there. In fact, Black populations appear in BRA and United States Census Bureau studies now to be growing much more rapidly in Matta-pan, Dorchester, Jamaica Plain, and Hyde Park than in Roxbury. None of this differential growth is reflected in the Camayd-Freixas methodology with its reliance on 1980 identifications of racial locations and its lack of hard data about current intracity migration by race. In brief, the ward and precinct identification of racial growth by Camayd-Freixas is ultimately intuitive and speculative. —2— The Republican plaintiffs are apparently willing to accept the total population figure for Boston as estimated by the DCC. See note 13 supra. However, they argue that the intracity population estimates are so speculative as to be constitutionally unacceptable. This theory, developed on the eve of trial in this matter, is apparently designed to justify their ultimate remedial request for at-large electoral districts in Boston. The Republican plaintiffs’ alternative is ultimately no alternative at all. It rejects all available data and seeks a suspect remedy. On both grounds it is unacceptable. The Republican plaintiffs assert that short of a state-wide redistricting based on 1980 federal census data, a city-wide multi-member district will be the only practicable way to assure equal representation in the Boston state legislative and city districts. Republican Plaintiffs’ Trial Memorandum at 7. But the prospect of resurrecting 1980 federal census data as a basis for redistricting is an idle one. The Supreme Court has been realistic about the limitations of federal census figures: These figures may be as accurate as such immense undertakings can be, but they are inherently less than absolutely accurate. Those who know about such things recognize this fact.... [I]t must be recognized that total population, even if absolutely accurate as to each district when counted, is nevertheless not a talismanic measure of the weight of a person’s vote under a later adopted reapportionment plan. The United States census is more of an event than a process. It measures population at only a single instant in time. District populations are constantly changing, often at different rates in either direction, up or down. Substantial differentials in population growth rates are striking and well-known phenomena. Gaffney v. Cummings, 412 U.S. 735, 745-46, 93 S.Ct. 2321, 2327-28, 37 L.Ed.2d 298 (1973). No party — not even the Republican plaintiffs — seriously suggests that the 1980 federal census provides a more current or accurate basis for calculating 1985 Boston population than does the DCC Report. In affirming Chief Judge Caffrey’s 1983 determination that the 1980 federal census was superior to the 1975 state figures, the First Circuit was careful to note that it could “find nothing in Chief Judge Caf-frey’s opinion to suggest that accurate State Census figures may not be used in later years to re-apportion the districts.” Latino Political Action Comm., Inc. v. City of Boston, 716 F.2d 68, 69 n. 1 (1st Cir.1983). The validity of this observation was underscored when a later three judge court, of which Chief Judge Caffrey was a member, indicated a disposition to defer to the 1985 state census as a basis for mid-decade redistricting in Massachusetts. MacGovern v. Connolly, 637 F.Supp. at 115. The willingness to accept a state’s alternative mechanism for determining population is a well-settled one. As the Sixth Circuit has observed, “[i]t is evident,” from the development of reapportionment case law, “that states may use adjusted population figures when redistricting between census years.” Young v. Klutznick, 652 F.2d 617, 624 (6th Cir.1981), cert. denied, 455 U.S. 939, 102 S.Ct. 1430, 71 L.Ed.2d 650 (1982). The Supreme Court emphasized at the outset of the development of reapportionment case law that “the Equal Protection Clause does not require the States to use total population figures derived from the federal census as the standard by which ... sub-stantial population equivalency is to be measured.” Burns v. Richardson, 384 U.S. 73, 91, 86 S.Ct. 1286, 1296, 16 L.Ed.2d 376 (1966). Estimates manifesting improved accuracy over outdated federal census have been both solicited, Wells v. Rockefeller, 273 F.Supp. 984, 991-92 (S.D.N.Y.), aff'd, 389 U.S. 421, 88 S.Ct. 578, 19 L.Ed.2d 651 (1967), and termed proper, Exon v. Tiemann, 279 F.Supp. 603, 608 (D.Neb.1967) by courts evaluating late and mid-decade reapportionment efforts. The requirement is that the adjustment figures be “thoroughly documented” and applied in a “systematic ... manner.” Kirkpatrick v. Preisler, 394 U.S. 526, 535, 89 S.Ct. 1225, 1231, 22 L.Ed.2d 519 (1969). That is clearly the case with the Decennial Census Commission figures. These figures may be challenged on methodological grounds. They may have their shortcomings. But the DCC figures were generated by a systematic, thoroughly documented and professional undertaking to obtain the most accurate estimate of Boston’s 1985 population as distributed through the City. In the absence of an alternative population distribution, the Republican plaintiffs offer a disfavored contrivance, the multi-member district. We recognize that there is some evidence of acquiescence by lower courts in the use of multimember districts. This has occurred where courts have found it necessary to employ multimember districts as an interim measure to deal with insurmountable difficulties in fashioning more comprehensive single member district remedies in a timely and non-disruptive fashion. See, e.g., Connor v. Finch, 431 U.S. 407, 415, 97 S.Ct. 1828, 1834, 52 L.Ed.2d 465 (1977); Chapman v. Meier, 420 U.S. at 21, 95 S.Ct. at 763; Mahan v. Howell, 410 U.S. at 334-35, 93 S.Ct. at 990 (Brennan, J. dissenting). We recognize as well that the use of at-large voting districts in Boston has been found not unconstitutionally discriminatory against Black voters. Black Voters v. McDonough, 421 F.Supp. 165 (D.Mass.1976), aff'd, 565 F.2d 1 (1st Cir.1977). But those cases that have evidenced unavoidable interim remedial use of multi-member districts have served to emphasize the Supreme Court’s conclusion “that single-member districts are to be preferred in court-ordered reapportionment plans unless the court can articulate a ‘singular combination of unique factors’ that justifies a different result.” Connor v. Finch, 431 U.S. at 415, 97 S.Ct. at 1834, citing Mahan v. Howell, 410 U.S. at 333, 93 S.Ct. at 989; Chapman v. Meier, 420 U.S. at 21, 95 S.Ct. at 763; East Carroll Parish School Board v. Marshall, 424 U.S. 636, 96 S.Ct. 1083, 47 L.Ed.2d 296 (1976). We find no such combination of unique factors here. Indeed to impose multi-member dis-tricting would be a repudiation of recent decisions made by the Boston electorate to abandon at-large arrangements and create single member districting. See generally Latino Political Action Committee, Inc. v. Boston, 568 F.Supp. at 1014. We do not deprecate the force of the criticism of the DCC methodology which the Republican plaintiffs have adduced through their experts. But that criticism, without a demonstration of an acceptable available alternative, is an insufficient ground upon which to base recourse to the discredited expedient of multi-member districts pursuant to a court ordered plan. Moreover, that criticism fails to convince us that the DCC figures are anything other than the most current and accurate measures of the distribution of Boston’s population. -G- Our exposure to the evidence in this case causes us to share the views expressed by Judge Sprizzo in his recent decision deferring to the United States Bureau of Census on questions of the use of statistical estimates. [T]he extensive testimony at trial overwhelmingly demonstrates that the determination as to whether the use of currently available adjustment techniques will provide a more or less reliable estimate of the population than the unadjusted census is an extraordinarily technical one, about which reasonable statisticians and demographers can and do disagree. Cuomo v. Baldridge, 674 F.Supp. 1089, 1105 (S.D.N.Y.1987). We recognize the force of the conclusion that no statistical apportionment of population can ever hope to be wholly accurate. But we also recognize that state redistricting authorities must make real-world decisions about which of the available competing plans best achieves the constitutional goals of equality of representation. Having reviewed the disagreements presented here and the alternatives available, we return to the DCC population figures. We have found no evidence that minorities were undercounted by the use of the DCC methodology. We are satisfied that there were no incentives to under-counting of minorities generally in the City of Boston census process and we do not find it unreasonable to rely on that process to identify relative growth within the wards of the City. The identified shortcomings of the Boston census appear to be related to the incentives encouraging an overcount in gross rather than on any systematic failure to identify minority persons. The DCC methodology avoids the problem of gross overcount and appears, in its use of Boston census figures, to have been a reasonable effort to capture the dynamics of interward/ growth. Coincidentally, if the intuitive and speculative judgments about racial growth rates which lay at the foundation of the Ca-mayd-Freixas methodology offered by the Political Task Force plaintiffs are accurate, the DCC allocation will reflect racial growth more successfully, than does the plaintiffs’ own approach. For example, rather than placing inordinate Black growth — as the Camayd-Freixas method does — in Roxbury, where it appears Black growth has not occurred the DCC figures are likely to identify growth in the wards where in fact it has taken place. If there is, in fact, relative minority growth, it has been identified and located under the DCC methodology in the ward in which it occurs. We find that although there are shortcomings in the DCC population allocations, the methodology adopted by the DCC was reasonable, that it was made without any racial bias or animus, and that it more closely approximates a proper determination of population allocations than any other methodology that has been made available to this court. The DCC methodology, whatever the weaknesses may be, is significantly more accurate than the alternative method of Dr. Camayd-Freixas. The DCC process and the persons directing and offering expert opinions to defend it appear to be without any particular bias on issues of race. Rather they appear to be disinterested demographers. We need not determine, however, whether a court would be constrained to accept the DCC Report out of considerations of deference. We find after careful and meticulous scrutiny that the systematic and thoroughly documented estimate provided by the DCC constitutes the most accurate current measure of the population in the City of Boston as of January 1, 1985. In short, the 1985 Decennial Census meets Federal constitutional standards for use by Massachusetts redistricting authorities. IV Under the Decennial Census Commission’s population figures — which we have determined in Part III to be an acceptable basis for the current readjustment of Massachusetts electoral districting — the ideal City of Boston electoral district would be 66,267 persons, the ideal state Senate district would be 143,661 persons, the ideal Executive Councillor district [each of which consists of five Senate districts] would be 718,305 persons and the ideal state House district would be 35,915 persons. The DCC figures were used by city and state redistricting authorities to create new districts for those electoral districts. The new City of Boston electoral districts and the State Senate and Council-lor Districts fall within the de minimis ten percent deviation rule that insulates state redistricting plans from effective challenge as violative of the “one person one vote” precept. The House districts, however, show more substantial deviations and consequently require an extended discussion. -A- As we have noted, see note 11 supra and accompanying text, in state districting — unlike federal congressional districting— precedent supports the treatment of a maximum population deviation of ten percent as de minimis and thus constitutionally permissible. Brown v. Thomson, 462 U.S. at 842-43, 103 S.Ct. at 2696. “A plan with larger disparities in population, however, creates a prima facie case of discrimination and therefore must be justified by the state.” Id. Above the ten percent ceiling, a 16.44 percent deviation has been found acceptable but that “percentage may well approach tolerable limits....” Mahan v. Howell, 410 U.S. at 329, 93 S.Ct. at 987. In any event, once the mathematical deviations exceed the prima facie level, [t]he inquiry then becomes whether it can reasonably be said that the state policy urged by [the state] to justify the divergences in the legislative reapportionment plan of the House is, indeed, furthered by the plan adopted by the legislature, and whether, if so justified, the divergences are also within tolerable limits. Id. at 326. As will appear in Part B infra, the divergences here extend well beyond the prima facie standard. Consequently, we examine in Part C infra, the fit between the state policies and their implementation to determine whether the state can justify the substantial departure from the one person one vote principle in its House districts. We conclude in Part D, infra, that the departures in the 1987 House Plan from principles of électoral equality cannot be justified and, consequently, are unconstitutional. -B- In framing the issues for trial the parties focused upon the deviation between the largest Boston District and the smallest House district elsewhere in the Commonwealth. The deviation between them is 18.31 percent, reflecting the fact that the Ninth Suffolk [Boston] House district is 7.76 percent over the ideal of 35,915 inhabitants and the Twelfth Norfolk District is 10.55 percent below. But in focusing the question, we are not free to wrench the issues out of their full context. The Supreme Court has been clear that: Regardless of possible concessions made by the parties and the scope of the consideration of the courts below, in reviewing a state legislative apportionment case this Court must of necessity consider the challenged scheme as a whole in determining whether the particular State’s apportionment plan, in its entirety, meets federal constitutional requisites. Maryland Committee for Fair Representation v. Tawes, 377 U.S. 656, 673, 84 S.Ct. 1429, 1438-39, 12 L.Ed.2d 595 (1964). Similarly, the Court has given lower courts firm instructions for evaluating the constitutionality of a state legislative apportionment scheme, regardless of what matters were raised by the parties_ [I]n determining whether a good faith effort to establish districts substantially equal in population has been made, a court must necessarily consider a State’s apportionment scheme as a whole. Only after evaluation of an apportionment in its totality can a court determine whether there has been sufficient compliance with the requisites of the Equal Protection Clause. Lucas v. Colorado General Assembly, 377 U.S. 713, 735 n. 27, 84 S.Ct. 1459, 1473 n. 27, 12 L.Ed.2d 632 (1964). When viewed in context, three salient features of the state legislative reapportionment relevant to the “one person one vote” principle appear. The House Redistricting Plan adopted shows: (1) a number of similar — indeed, even more extreme — deviations from the prima facie ten percent standard demonstrated by pairing the Ninth Suffolk and the Twelfth Norfolk Districts; (2) deviations which are pervasive throughout the House plan; and (3) deviations which affect a substantial percentage of Massachusetts voters. (1) While higher than any deviation— with one unique exception (see generally the discussion of Brown v. Thomson, Part D infra — ever permitted by the Supreme Court, the deviation between the largest Boston district and the smallest district elsewhere in the Commonwealth is in fact not the largest deviation in the 1987 Massachusetts House Plan. The most populous House district, the Seventeenth Middlesex, when combined with the least populous district, the Twelfth Norfolk, yields a variance of 21.9 percent. Moreover, using the Seventeenth Middlesex and the Twelfth Norfolk as partners for illustrative pairings, a substantial number of deviations larger than any — again apart from the one highly unique exception — ever accepted by the Supreme Court can be assembled from the House Plan. For example, if the Seventeenth Middlesex is paired with the 28 smallest House districts every pairing exceeds 16.9 percent. By the same token if the Twelfth Norfolk district is paired with the twenty-seven largest House districts every pairing exceeds 16.5 percent. Nor are these inordinate pairings the result of statistical anomalies at both ends of the spectrum. Any of the thirteen most populous House districts can be compared with any of the thirteen least populous districts to create a population deviation higher than 16.44 percent. (2) The deviations from the norm occur throughout the state. In order to yield the inordinate pairings between the Seventeenth Middlesex and the twenty-eight smallest districts, districts from the counties of Barnstable, Berkshire, Bristol, Essex, Hampden, Norfolk, Plymouth, Worcester — and even the same county, Middlesex — are used. Similarly, in order to create inordinate pairings between the Twelfth Norfolk and the twenty-seven largest districts, districts from the counties of Barnstable, Bristol, Essex, Hampden, Mid-dlesex, Plymouth, Worcester — and, again, even the same county, Norfolk — are used. No party has contended, and we find nothing in the record to suggest, that the deviations have been designed to advantage a particular type of district, e.g. suburban, urban or rural; agricultural, industrial or maritime; or a particular geographical part of the Commonwealth. Indeed, within the City of Boston as a whole, which is the primary focus of the plaintiffs’ complaint, it appears that the deviations all but cancel each other out. (3) The inordinate divergences pervasive in the 1987 House Plan, while not apparently designed to advantage or disadvantage any particular element of the electorate, substantially affect large portions of the population. The districts which can create paired deviations in excess of any — save the one anomaly — approved by the Supreme Court in a state-wide reapportionment challenge, involve over one-third of the districts spread throughout the Commonwealth. The residents of those districts, a group that constitutes over one-third of the Commonwealth’s inhabitants, are consequently subject to extreme malap-portionment. Thus, we find the questionable deviations from one person one vote principles in the 1987 Massachusetts House Plan to be inordinate, pervasive and to have a direct effect on the relative electoral power of a substantial percentage of the Commonwealth’s voters. We turn to the policy justifications offered for this failure to comply with the fundamental directive of Reynolds v. Sims and its progeny. In doing so, we will be inquiring whether those justifications have been implemented with consistency in the 1987 House plan and what history tells us about the force and thrust of those justifications in Massachusetts. -C- The basic state policies governing the division of the Commonwealth into 160 single member House districts are set forth in the Massachusetts Constitution. Amendment Art. Cl, § 1, provides that those districts shall be of contiguous territory so that each representative will represent an equal number of inhabitants, as nearly as may be; and such districts shall be formed, as nearly as may be, without uniting two counties or parts of two or more counties, two towns or parts of two or more towns, two cities or parts of two or more cities, or a city and a town, or parts of cities and towns into one district. Such districts shall also be so formed that no town containing less than twenty-five hundred inhabitants according to said census shall be divided. The definitive construction of those policies was elaborated upon by the 1987 House Joint Special Committee: Avoidance of uniting counties was described “more as a constitutionally-endorsed policy than as an iron-bound mandate.” 1987 House Joint Special Committee Report at 24. Avoidance of uniting or dividing communities was said, “as [with the] ‘county limitation’ [to be] regarded ... as a strong constitutional policy declaration rather than as an absolute mandate.” Id. at 25. Special protection for smaller communities was said to indicate an “intent that the General Court be authorized to divide larger towns, if need be, to meet the population-equality standard notwithstanding the above policy against the ‘fragmentation’ of municipalities.” Id. at 25. In implementing these state constitutional principles, the 1987 House Joint Special Committee adopted a series of guidelines earlier enunciated by the 1977 House Joint Special Committee Report. These guidelines directed: (a) the creation of districts within the boundaries of very populous municipalities, (b) legislative district boundaries coterminous with municipal boundaries of cities or towns populous enough to compose their own single representative district, (c) the grouping together of whole smaller communities in single representative districts, (d) minimal divisions of towns of 6,000 or more inhabitants, and of cities, in ways which would deprive them of a distinctive voice in the House of Representatives, and (e) no division of any town of fewer than 6,000 inhabitants. Id. at 30. These various rearticulations of the state policies found in the record of this case may be synthesized into a hierarchy of policy imperatives: 1. State constitutional mandates: (a) single member districts, (b) contiguity of territory, (c) protection of single representation of small municipalities, defined as having fewer than 2,500 inhabitants; 2. State constitutional directives not amounting to mandates: (a) avoidance of the uniting of counties; (b) avoidance of the uniting and dividing of municipalities; 3. Glosses on the constitutional policies provided by legislative committees: (a) guidelines for the establishment of districts involving communities larger than those having 6,000 inhabitants and (b) definition of smaller communities requiring special protection as those communities having 6,000 or fewer inhabitants. Out of this proliferation of rearticula-tions, it is evident that while most of the state’s policies easily can be observed under virtually any redistricting plan, the key impediment to the equality of House districts is the professed desire to embody respect for subordinate political boundaries in creating districts. But this professed desire has not been strictly honored in current practice nor does it provide an adequate justification for the inordinate, pervasive and substantial departures from equality in districting which are evidenced in the House Plan. -1- The 1987 House Plan itself shows no rigorous adherence to a policy of respect for the boundaries of political subdivisions. County boundaries are systematically disregarded and municipal boundaries are not considered a systemic constraint when necessity and convenience suggest otherwise. The proscription against uniting counties in districts — which stands with the policy of respect for municipalities at the second level of the hierarchy of Massachusetts public policies applicable to reapportionment decisions, that is, as a directive but not a constitutional mandate — is honored only in the breach. House districts cross county lines in every county save the singular Cape Cod and Islands region of the state (Barnstable, Duke and Nantucket counties) and the far west (Berkshire County). Under the House Plan, twenty-three cities and nine towns are divided by the 1987 redistricting. Simple mathematics necessitated that most of these communities would be divided because the number of inhabitants substantially exceeded the ideal House district size of 35,915 inhabitants. But a number of smaller communities are also fragmented under the plan. The 14,-522 inhabitants of Foxborough, the 28,435 inhabitants of Randolph, the 21,836 inhabitants of Stoneham, the 13,174 inhabitants of Westwood and the 17,704 inhabitants of Wilmington are in the case of each town split between two members of the House of Representatives. It is apparent that to meet the practical imperatives of districting, Massachusetts is prepared to unite its counties, to unite various of its communities and to fragment its larger — and some of its smaller — municipalities. The policy of respect for political subdivision boundaries then, both in articulation and in practice, is heavily seasoned with a healthy recognition of competing considerations. Among those competing considerations, the most important historically in Massachusetts, at least since 1857, has been electoral equality. When we approach the public policies of the Commonwealth of Massachusetts which are called into play and the relative significance of each, we recognize that there are few more forceful statements of the preeminence of electoral equality than Chief Justice Rugg’s observations for the Massachusetts Supreme Judicial Court in 1916 regarding Amendment Article XXI of the Massachusetts Constitution, enacted in 1857: The great principle established by this amendment is equality of representation among all the voters of the Commonwealth. That is a fundamental principle of representative government. It is preeminently so in a State whose Declaration of Rights declares that ‘all men are born free and equal,’ and in a nation whose Declaration of Independence asserts that ‘all men are created equal.’ There can be no equality among citizens if the vote of one counts for considerably more than that of another in electing public officers. The true spirit and meaning of the Constitution is that each voter has an equal voice in the enactment of laws and in the election of officers of the State. Such equality must be secured in all laws for the choosing of representatives in the General Court or the Constitution is violated_ Inequality of representation apparent on the face of the apportionment offends against this constitutional provision, a provision of the very essence of any conception of equality among voters, each of whom is the peer in political right of every other. Attorney General v. Suffolk County Apportionment Commissioners, 224 Mass. 598, 604, 113 N.E. 581 (1916). Amendment Article XXI, which was the predecessor of Amendment Article Cl, “was designed,” as the Justices of the Supreme Judicial Court under Chief Justice Lemuel Shaw found 130 years ago in an opinion rendered in response to a question regarding Amendment Article XXI from the House of Representatives, “to supersede and replace the long practice of electing representatives through the medium of town organizations, and the agency of municipal officers_” Opinion of the Justices, 76 Mass. (10 Gray) 613, 615 (1858). “The great object to be attained [by Amendment Article XXI] was ... in conformity with the theory of representation, to secure as nearly as possible an equality in the ratio of representatives and legal voters throughout the Commonwealth.” Id. at 615-16. Thus as early as 1857 Massachusetts had abandoned reliance on town organization to structure legislative districts in favor of a system whose preeminent goal was to achieve the greatest possible equality among voters. Municipal boundaries remained a consideration in the structuring of legislative districts, but those boundaries became over time a subordinate concern until by 1987, in words adapted from the 1977 House Joint Committee Report, respect for municipal boundaries had come to be viewed “as a strong constitutional policy declaration rather than as an absolute mandate.” 1987 House Joint Special Committee Report at 25. In light of this history, the reference in the 1987 Joint Special Committee Report to “keeping communities intact in the redistricting process [as] ... an important consideration in the redistricting of the House of Representatives in order to afford the best achievable forum for these local voices ...” id. at 30, must be given a narrow reading. Moreover, in terms of the hierarchy of imperatives among the Massachusetts public policies, it must be emphasized that the general anti-fragmentation policy is simply a directive and is not a mandate. There is, of course, a constitutional anti-fragmentation mandate in Massachusetts but that is only designed to prevent fragmentation of the smallest communities; no such mandate attaches to the larger communities. This carefully drawn distinction in the Massachusetts Constitution is an understandable one. A community must have a critical mass in order to have its local voice heard within a legislative district. Fragmentation of smaller communities impairs their ability to reach this critical mass. The point at which protection of the communities from non-fragmentation is mandated has been explicitly addressed by Massachusetts. It is defined as 2500 inhabitants by the Massachusetts Constitution and 6,000 inhabitants by the 1977 and 1978 House Joint Special Committees. A community above that size is no longer entitled, as a matter of Massachusetts constitutional public policy, to specially mandated protections against fragmentation, as the experience of Foxborough, Randolph, Stoneham, Westwood and Wilmington under the 1987 House plan demonstrates. There are good reasons why some communities might welcome division into two legislative districts. So long as the municipal voice is heard, it may, for instance, be of significant value to municipal interests to have two or more members of the house who have a special concern for the municipality as a constituency. In any event, beyond what is mandated by the Massachusetts Constitution for the protection of smaller communities, we find no imperatives in the public policy of Massachusetts requiring that municipal boundaries of the larger Massachusetts cities and towns must be adhered to in the formation of House districts in a fashion sufficient to overcome the application of ordinary one person one vote principles. -D- Nothing in reapportionment case law supports the wholesale departures from principles of one person one vote embodied in the 1987 House Plan. As a litigant in this case, the Commonwealth looks principally to Brown v. Thomson, 462 U.S. 835, 103 S.Ct. 2690, 77 L.Ed.2d 214 (1983), and Mahan v. Howell, 410 U.S. 315, 93 S.Ct. 979, 35 L.Ed.2d 320 (1973), for authority. But far from justifying the departures of the 1987 House Plan, Brown and Mahan both illustrate why such departures cannot be countenanced in Massachusetts. Brown v. Thomson presented what Professor Tribe accurately calls a “strange set of facts.” L. Tribe, American Constitutional Law at 1072-73 n. 10. Niobrara County, Wyoming, an isolated rural county of only 2,924 persons was given a House seat despite the fact that the ideal House district would have contained 7,337 persons. The Supreme Court, which adhered to the parties’ decision to consider only the deviation created by the Niobrara County anomaly, found that the “case presents an unusually strong example of an apportionment plan the population variations of which are entirely the result of the consistent and non-discriminatory application of a legitimate state policy.” Brown v. Thomson, 462 U.S. at 844, 103 S.Ct. at 2697. The Wyoming policy of providing each county with at least one House seat of its own was based on the state constitution and was followed consistently throughout the state in the decades since statehood. Id. at 843, 103 S.Ct. at 2696. Moreover, the policy recognized the peculiar size and population distribution of the state and a governing structure for the state in which “the counties are the primary administrative agencies of the State government.” Id. at 841 n. 5, 103 S.Ct. at 2695 n. 5 (citing district court opinion). Unlike the firm and historically consistent constitutional mandate reflecting peculiarities of size and distribution of the population within the state, to which deference was shown in Brown, constitutional directives in Massachusetts to respect the boundaries of political subdivisions in the formation of electoral districts have, as we have observed, tended to be made applicable only as convenient and then with substantial elasticity. The proscription against uniting political subdivisions — which in Massachusetts stands, as a directive and not a mandate, on an equal constitutional footing with the general nonfragmentation policy for larger municipalities — was the specific state policy at issue in Brown. But that policy was rejected by Massachusetts in a setting not unlike the one addressed in Brown. The closest analogy to the unique demographic setting of Niobrara County, Wyoming which Massachusetts has to offer— the underpopulated insular counties of Nantucket and Dukes (Martha’s Vineyard and the Elizabeth Islands) — ceased in 1977 to receive special consideration in the arrangement of House districts. This was done in recognition of the preeminence of population equality in reapportionment. The 1977 Joint Special Committee came to the conclusion “that it is now constitutionally impermissible for the General Court to tailormake either a single representative district for each island county, or a single such district embracing both those counties.” House No. 5900 at 70. The 1977 Joint Committee could foresee no particular dangers to evolving Massachusetts policy by the consequent disregard of the uniqueness of the island communities. “Since the abandonment of town-based representation in the General Court in 1857, a revival of which would not be countenanced under the 'one man one vote’ criteria of the Federal Constitution, many small mainland communities have been long grouped together in representative and senatorial districts, without anyone’s suggesting seriously that this deprived their inhabitants of representation in the General Court.” Id. at 71. Nor do we confront here the consistent historical Virginia state policy of maintaining the integrity of political subdivision lines found in Mahan. Even if Massachusetts could show historically consistent adherence to a policy of respect for political subdivision boundaries and could further show, as it cannot, that its adherence to that policy in the 1987 House Plan produces the minimum divergence from the goal of electoral equality, Mahan stands as a reminder that such divergences must also be “within tolerable limits.” 410 U.S. at 326, 93 S.Ct. at 986. Apart from the inapplicable anomaly presented by Brown, the case law is in the same position now as it was when reviewed by the three judge court which struck down a reapportionment plan in Hawaii involving a maximum 16.02 percent deviation among House districts. Although noting that it was finding the Hawaiian House Plan unconstitutional because it failed reasonably to further the state’s expressed policy of distributing House seats equally among the Islands, the Travis court observed: as far as we can ascertain from independent research, it appears that no court since Mahan has upheld a legislative reapportionment scheme which includes a maximum deviation of 16.5% or more. Apparently most courts have read the Supreme Court’s caveat in Mahan as implicitly setting this figure as the outermost limit allowed by the Constitution. Id. at 562 and n. 19 (D.Haw.1982) (footnote deleted). Court created plans — which, to be sure, are held to higher standards than state created plans — containing