United States v. Green

389 F. Supp. 2d 29, 2005 U.S. Dist. LEXIS 19037, 2005 WL 2109114

Opinion

MEMORANDUM AND ORDER RE: DEFENDANTS’ CHALLENGE TO THE COMPOSITION OF THE JURY VENIRE

GERTNER, District Judge.

TABLE OF CONTENTS

I.INTRODUCTION.35

II. BACKGROUND.40

A. Step One: Choice of a Federal Forum — from 20% African-American Representation in Suffolk County to 7% African-American Representa-

tion in the Eastern Division of the District of Massachusetts.40

B. Steps Two Through Four: From 7% African-American Representation in the Eastern Division to 3% on the Available Jury Wheel .42

1. Step Two: From Source Lists to the Master Jury Wheel.42 2. Steps Three and Four: Determining the “Available Pool” or the

“Qualified Wheel” .43

C. Steps Five and Six: Jury Impanelment — Venire to Petit Jury Selection from 3% to Nil.45

D. History of Minority Underrepresentation in Massachusetts and Related Federal Litigation.46

1. The 1993 Boston Litigation.46

2. 1994 Supreme Judicial Court Gender and Race Bias Report.47

3. Federal Litigation Between 1984 and 1999 .'.47

E. Defendants’ Case.47

III. DISCUSSION....,. ^ fQ

A. Fair Cross-Section Challenge . tn

1. Fair Cross-Section Framework. w

a. Second Prong: Underrepresentation. cn i — i

(1) By What Statistical Means Shall the Exclusion of Members of a Cognizable Group Be Measured?. cn v-i

(2) How Much Exclusion of Members of a Cognizable Group Is Significant for Constitutional Purposes?. or

b. Third Prong: Systematic Exclusion. or en

(1) Is the Disparity Found under the Second Prong Caused by Happenstance (Which Is Not Actionable) or Is it Caused by Official Action or Inaction of Some Sort (Which May Be Actionable)?. Ol cn

(2) Even if Official Misfeasance Contributes Somewhat to the Disparity in Representation, Do the Defendants Have to Show Precisely How Much of the Disparity Is Attributable to Such Factors?. cn CJi

c. Hybrid Approach. cn -O

2. Defendants’ Case. Or

a. First Prong: Distinctiveness. cn

b. Second Prong: Underrepresentation. en —3

c. Third Prong: Systematic Exclusion. Or CO

(1) Shortcomings of The Resident Lists — Undercounting and Overcounting. cn co

(2) Shortcomings of Summonsing — Demographics, Logistics and Nonresponses. ©

(a) Undeliverables. 2

b. Section 1863(b)(2)’s Duty to Supplement.

(b) Nonrespondents. rH ZD

(3) Inactive Voter Lists . Cm ZD

3. Conclusion: Defendants Have Not Made Out A Constitutional Violation.

B. Statutory Challenge.

1. The JSSA’s Proportionality Requirement.

to Comply” with the JSSA.

a t Act?_

(1) The Statutory Language.

(2) Legislative History.

(3) The District of Massachusetts Resident List Exception.

c. Defendants Proven a “Failure to Comply” Without Supplementation?. So

Failure to Supplement the Resident Lists Amount to a Substantial Statutory Violation?. CO Cr-

e. Supervisory Powers. ^ t-

ID Cr-

g. Afterward: The Government’s Objections to the Proposed Remedy.

IV. CONCLUSION. .79

I. INTRODUCTION 1

Darryl Green (“Green”) and Branden Morris (“Morris”) are African-American men who are likely to be tried before all white, or largely white, juries. Such an outcome should be profoundly troubling, to say the least. Indeed, the District of Massachusetts has wrung its collective hands over the problem of minority underrepre-sentation on its juries for over a decade. However significant the lament before, the prospect is uniquely chilling here: Green and Morris face the death penalty. Their all white, or largely white, juries could well decide whether they will live or die.

Morris and Green, along with three co-defendants, 2 are charged with participating in a racketeering enterprise — the “Es-mond Street Posse” — through which they allegedly sold crack cocaine and marijuana, protected their sales turf, and carried on a violent dispute with a rival gang. That dispute led to a number of murders and attempted murders during 2000 and 2001. The death of Terrell Gethers prompted the government to charge Morris and Green with murder in aid of racketeering under 18 U.S.C. § 1959(a)(1) and to seek the death penalty against them.

Defendants claim that the racial compo sition of the jury wheel 3 for the Eastern Division of the District of Massachusetts (“Eastern Division”) violates the Sixth Amendment and the Jury Selection and Service Act, 28 U.S.C. § 1861 et seq. (the “Act” or “JSSA”). 4 They allege that the federal officials use state resident lists that are inaccurate and out of date, particularly from the cities and towns with the highest percentage of African-Americans. They move to dismiss the charges against them, or, in the alternative, to stay the case until a jury can be assembled that comports with the Constitution and the JSSA. 5

Defendants’ claims are ironic: Massachusetts pioneered the use of resident lists in place of voting lists for jury selection precisely to maximize minority participation. But the duty to prepare and update these lists has remained an unfunded mandate, fulfilled with varying success across the District. According to defendants, the more affluent and whiter communities can afford to properly maintain the lists; the poorer 6 , more racially diverse communities cannot. Put simply, an Eastern Division resident has a better chance of getting on a jury if she hales from more racially and economically homogenous towns like Needham or Dover, than if she is from more racially and economically diverse towns like Lynn, Brock-ton or New Bedford. Residents of heavily African-American, poor, and urban communities, like Roxbury and Dorchester, may fare even worse than those from the latter towns.

The government opposes defendants’ motion. It defends the lawfulness of its procedures for compiling the jury wheel, and, in effect, the overwhelmingly white juries those procedures are likely to yield. It denies that there is any official misfeasance, federal or state, or that misfeasance is responsible for the underrepresentation of African-Americans. The government argues that the reason for the underrepre-sentation is that substantial numbers of African-Americans choose not to return court questionnaires and that residents of poorer and heavily minority communities are so transient that their addresses are not easily captured on any resident list.

The stakes could not be higher. Undermining the right to a representative jury casts a pall over all jury trials in our District. The issue is particularly important for the capital jury, not only because of the stakes, but also because of that jury’s unique role. It renders not simply a factual judgment — guilt or innocence — but “an ethical judgment expressing the conscience of the community.” Jeffrey Abramson, Death-Is-Different: Jurisprudence and the Role of the Capital Jury, 2 Ohio St. J.Crim. L. 117, 119 (2004) (citing Spaziano v. Florida, 468 U.S. 447, 469, 104 S.Ct. 3154, 82 L.Ed.2d 340 (1984) (Stevens, J., concurring in part and dissenting in part)). 7 And, as Justice Marshall eloquently noted, “[w]hen any large and identifiable segment of the community is excluded ... the effect is to remove from the jury room qualities of human nature and varieties of human experience, the range of which is unknown and perhaps unknowable.” Peters v. Kiff, 407 U.S. 493, 503, 92 S.Ct. 2163, 33 L.Ed.2d 83 (1972) (Marshall, J.). 8 The result is not merely the appearance of bias; it may well be its reality. Id.

Defendants do not raise the first constitutional challenge to the racial composition of the federal venire. There have been many such challenges, all unsuccessful, largely because of the rigorous standards imposed by the courts, including the First Circuit. While others have criticized those standards, including judges on this Court, I have no choice but to apply them. In doing so, I conclude that defendants have not established a constitutional violation.

But my analysis does not end here. The Constitution provides a floor, not a ceiling, to the Court’s obligation to provide representative juries. The JSSA imposes higher standards on public officials to supplement their source lists when necessary to “foster the policy and protect the rights” to a fair cross-section jury selection process. 28 U.S.C. § 1863(b)(2). I find that those standards have been violated here.

In brief, my findings are:

All, or nearly all, white juries are made much more likely by a single decision of the Executive: The United States Attorney’s office has opted to prosecute “street crime” in federal court, rather than in the courts of the Commonwealth of Massachusetts. With that decision, the available pool of African-American jurors plummets from 20% in Suffolk County, where defendants’ alleged crimes took place, to roughly 7% in the Eastern District of Massachusetts. And the punishment escalates from life imprisonment in the state courts to the death penalty in the federal courts. No matter how troubling the impact, the law gives the federal prosecutor the right to make this decision.

Making matters even worse, however, the 7% African-American representation is diluted further before a single juror is sworn in federal court. African-American representation plummets to roughly 3% or less in the Eastern Division jury pool after jury summonses are returned, at least in part because of outdated and inaccurate resident lists. As a result, the vast majority of Eastern District juries will not have a single African-American member.

As striking as it is, however, this data is still not enough to make out a prima facie case of a Sixth Amendment violation, at least under current standards. Defendants not only have to show that African-Americans are underrepresented in the jury pool in relation to their numbers in the population, they also have to show that the underrepresentation has reached a certain threshold percentage. That threshold percentage has been set so high by First Circuit precedent that it is virtually unreachable in this District. And, even if defendants prove underrepresentation of a certain degree, they must also show how that underrepresentation occurs — that it is attributable to something systemic, like official action, rather than happenstance. Finally, even if they show that official misfeasance contributes to some degree to the problem but they cannot show precisely how much, their claim may still fail.

The constitutional question should not simply be about numbers. The real question is: How much underrepresentation of African-Americans is constitutionally tolerable in a system that promises a representative jury? Does it matter that the choice of forum — the Executive’s choice— has already affected the nature of the deci-sionmaker, diluting the jury pool from 20% to 7% African-American? When the federal government opts to prosecute street crime, when it seeks extraordinary penalties, when a substantial percentage of those defendants are African-American and the overwhelming majority of jurors are white, perhaps even a 2 or 3% under-representation is far too much. Existing constitutional standards, however, say otherwise.

Since on this record defendants 1) cannot prove the magnitude of the disparity that the First Circuit has thus far required, although they have proved substantial disparity, and 2) cannot prove the precise extent to which that disparity is attributable to flaws in the system itself, although they have proved that official action and inaction contributes to the problem, I am obliged to deny their constitutional challenges.

The JSSA, however, sets the bar higher. The statute imposes an affirmative obligation on districts to use jury selection processes that ensure random selection from a “fair cross-section of the community.” 28 U.S.C. § 1863(b)(2). However, I find that the failure of the Court to direct the Federal Jury Administrator to supplement the existing flawed resident lists amounts to a statutory violation calling for remedial measures. Under the JSSA and my supervisory authority, I will order, among other things, that additional summonses be sent out in this case for each summons that is returned as “undeliverable,” signifying a bad address, as well as for each summons that is not responded to after a second mailing. I will also order that steps be taken to strike inaccurate addresses from the list, so that the same wrong addresses do not recur in case after case. The remedy is entirely justified, consistent with the statute, and the District’s Jury Plan as the Court’s expert 9 and the Chief Judge of this Court have concluded. 10

And I will go even further. My findings and the report of the court-appointed expert will be submitted to the appropriate court authorities for systematic district-wide attention. The goal is to make certain that everything that lawfully can be done to increase minority representation is done, including geographically-weighted mailings that take into account the historical data concerning jury response rates across the District, as described infra in Part III.B.2J. In addition, I will urge my state colleagues to address the fundamental problem- — an unfunded state mandate to produce accurate resident lists that is carried out inconsistently across the Commonwealth.

Substantial resources have already been devoted to litigating this issue. The parties filed voluminous briefs; the Court held several days of hearings. Defendants sent questionnaires to the clerks of all cities and towns comprising the Eastern Division to determine how they compile their jury lists. They were also given funds for a jury expert, Professor Andrew Beveridge of Queens College (“Bever-idge”). 11 Although the government did not hire an expert, the Court took the extraordinary step of appointing its own, Professor Jeffrey Abramson of Brandéis University (“Abramson”), pursuant to Fed. R.Evid. 706. 12

In the subsequent sections, I will first outline the background of the jury selection processes and the defendants’ case. I will then address the legal arguments, answering the following questions:

1. The constitutional fair cross-section guarantee: Whether the evidence presented by defendants establishes any constitutional violation of defendants’ right to have petit jurors chosen from a fair cross-section of the community (Section III.A.);

2. The statutory substantial proportionality requirement: Whether the processes for selecting jurors, as delineated in the District Court’s Amended Jury Plan, 13 violate the provision of the Jury Selection and Service Act requiring that political subdivisions within the Eastern Division be “substantially proportionally represented,” 28 U.S.C. § 1863(b)(3) (section III.B.l);

3. The statutory “substantial compliance” requirement: Whether the record establishes violations of the District Court’s Jury Plan or the Act, thereby entitling defendants to a remedy under 28 U.S.C. § 1867(d) for “substantial failure to comply with provisions” of the Act (Section III. B.2);

4. Supervisory jurisdiction: Even in the absence of specific findings under 1, 2 or 3 above, what steps can the Court take to address, at least in part, the problems revealed in this litigation (Section III.B.2.e).

One final note: In an earlier decision, I considered whether defendants should be tried before one jury determining liability and a second determining punishment. Only the punishment jury would be “death-qualified.” United States v. Green, 343 F.Supp.2d 23 (D.Mass.2004), as amended, 348 F.Supp.2d 1 (D.Mass.2004). I made this decision as a matter of case management, to avoid the complex jury selection process death-qualification requires at the liability stage. I was reversed. United States v. Green, 407 F.3d 434 (1st Cir.2005). I now ask the parties to revisit the issue in the context of the jury selection issues raised in this opinion. Death-qualification of the jury may well further diminish African-American jury representation in this District from roughly 3% to nil.

Thus, the following questions are raised: Whether the practice of death-qualification has a disproportionate impact on the already small numbers of African-Americans in the jury venire, whether this issue is cognizable under the Sixth Amendment, or the Due Process Clause of the Fifth Amendment, and whether there is another less burdensome means of accomplishing the government’s legitimate goal of seeking the death penalty before a jury that is amenable to such a punishment.

The data presented by defendants raises grave concerns. Action is not only called for but imperative. As the court-appointed expert concluded: “Metaphorically speaking, there has to be a statute of limitations on how long a District can lament the undesirability of the underrepre-sentation of minorities in its jury pools without feeling compelled to act with imagination to do better.” Abramson, Report at 64-65.

II. BACKGROUND

Before examining the extent and causes of African-American underrepresentation on District of Massachusetts juries, I begin with an overview of the jury selection process in our district. 14 The process involves six steps: 1) determination of federal district boundaries and selection of a forum (state/federal); 2) creation of a “master jury wheel” from resident lists; 3) random selection of potential jurors to receive summonses and questionnaires; 4) selection of the “available jury wheel” on the basis of questionnaire responses; 5) creation of the “jury venire” on the basis of qualified jurors who respond to the notice to appear; and 6) selection of the petit jury after voir dire and the peremptory challenge process.

A. Step One: Choice of a Federal Forum — from 20% African-American Representation in Suffolk County to 7% African-American Representation in the Eastern Division of the District of Massachusetts

Jury districts are created by statute, court rule, or both. The District of Massachusetts was created by federal statute; the Eastern Division was created by court rule. Pursuant to 28 U.S.C. § 1869(e) and the District of Massachusetts Jury Plan, 15 the District of Massachusetts is divided into three divisions for petit and grand jury selection — the Eastern, Central and Western Divisions. 16 While legislative districts are drawn with the representativeness of racial groups in mind, see Voting Rights Act, 42 U.S.C. § 1973, et seq., judicial districts are arbitrary, administrative contrivances.

Law enforcement chooses the forum (federal or state) for prosecution of a crime. 17 Taken together, administrative decisions with respect to district boundaries and the Executive’s choice of forum define the geographic areas within which potential jurors will reside, and what “representativeness” means in connection with jury pools drawn from those areas.

The substantive crimes with which defendants are charged — homicide and street-corner narcotics trafficking — have traditionally been prosecuted in state courts. Had this case been brought in state court, the “community” for the purpose of determining what comprises a “fair cross-section” would be Suffolk County; in federal court, the relevant community includes all of eastern Massachusetts. In Suffolk County, defendants’ juries would be drawn from a voting-age population that is roughly 20% African-American. In the Eastern Division of Massachusetts, only roughly 7% of the voting-age population is African-American. 18

Census data for Massachusetts, like most states, shows that minority populations are clustered in urban areas. By choosing federal court and thereby expanding the jury district to include the more racially homogenous suburbs, the government invariably dilutes minority' — ■ and even urban — representation in the pool from which defendants’ juries will be selected. While the Sixth Amendment demands representativeness, it does not require courts to second-guess the boundaries of the judicial district. Thus, when the government federalizes local crime in the more diverse cities of Lawrence, Lowell, or Boston, on this end of the state, or Springfield, on the other, it homogenizes the decisionmaker. 19 And the law allows it

to do so. 20

B. Steps Two Through Four: From 7% African-American Representation in the Eastern Division to 3% on the Available Jury Wheel

1. Step Two: From Source Lists to the Master Jury Wheel

The State Office of the Jury Commissioner (“OJC”) starts with a “source list” comprised of lists of names and addresses of potential jurors, and then randomly draws a percentage of the names to create a “master jury wheel” for the federal Jury Administrator. The JSSA defines the procedures for generating names for the master jury wheel, and also mandates the creation of a more specific district court jury selection plan. See 28 U.S.C. § 1861 et seq. Federal courts may draw the names of prospective jurors from either voter registration lists or the lists of actual voters within their districts. 28 U.S.C. § 1863(b)(2). But there is an alternative: Each federal district court “shall prescribe some other source or sources of names [of prospective jurors] in addition to voter lists where necessary to foster the policy and protect the rights secured by sections 1861 and 1862 of this title.” 21 Id.

A 1992 amendment to the Act specifically provided that the District of Massachusetts “may require the names of prospective jurors to be selected from the resident list provided for in chapter 234A, Massachusetts General Laws, or comparable authority, rather than from voter lists” as its source of names for the master jury wheel. 22 28 U.S.C. § 1863(b)(2). The Act was amended precisely because of serious concerns about the racial composition of jury pools drawn from voter lists. 23 See, e.g., United States v. Levasseur, 704 F.Supp. 1158, 1164 (D.Mass.1989). Minorities did not vote in the same proportion as did their white counterparts. See Bernard Grofman et ah, Drawing Effective Minority Districts: A Conceptual Framework and Some Empirical Evidence, 79 N.C. L.Rev. 1383, 1404 (2001) (citing Kimball Brace et al., Minority Voting Equality: The 65 Percent Rule in Theory and Practice, 10 Law & Pol’y 43, 47-48 (1988)). Using voter lists to compile juror lists effectively extended the gap in political participation between the races into the jury arena; resident lists represented a pioneering effort to produce jury wheels more closely reflecting the racial composition of the districts.

For Massachusetts, the OJC compiles a single-numbered statewide resident list from the resident lists allegedly prepared annually by every Massachusetts city and town. Every city and town is required under M.G.L. ch. 234A to make such a list of all residents who resided in the town as of each January. 24 Unfortunately, no state funds are appropriated to ensure that the statutory requirements are fulfilled. 25

To construct the Eastern Division’s master jury wheel, the OJC randomly draws 1% of the names on that portion of the OJC resident list that represents the 190 cities and towns in the Eastern Division. 26

2. Steps Three and Four: Determining the “Available Pool” or the “Qualified Wheel»

During step three, the federal Jury Administrator determines the approximate number of jurors needed in any given week or month according to the number of jury trials scheduled in the courthouse. He then randomly selects that number of potential jurors from the master jury wheel and mails them summonses and questionnaires. The questionnaire asks potential jurors several questions about their citizenship, occupation, and criminal history, principally to determine who is qualified to serve on a federal jury. 27 The questionnaire also asks potential jurors information about their demographic characteristics, like age and race.

During step four, potential jurors who complete and return questionnaires, and are not eligible for automatic disqualification, are placed in the “available jury wheel” — the pool of people who are qualified for federal jury service.

Significantly, returned questionnaires provide the first opportunity to measure the racial composition of the Eastern Division jury pool. 28 However, not all people who are mailed a summons and questionnaire return the questionnaire. 29 Some never receive the summons; others receive the summons but fail to respond by returning the questionnaire. Summonses marked “undeliverable” are clearly in the former category. “Nonresponses” — summonses sent out and never returned — may include both individuals who did not receive summonses, because addresses were wrong, and those who chose not to respond. The thrust of defendants’ challenge is that the master resident list is plagued with inaccurate names and addresses, inaccuracies that stem from the failures of officials in certain cities and towns to generate accurate lists in the first instance and/or update them.

Combining data from 2001 through 2003, approximately 88% of persons who returned their questionnaires identified their race. Despite some limitations in the data, 30 it is abundantly clear that African-Americans are persistently underrepresented in Eastern Division available juror pools. In 2001, African-Americans comprised 6.75% of the Eastern Division population according to United States Census (“U.S.Census”) estimates, but only 3.08% of those who identified race on returned questionnaires. 31 In 2002, African-Americans comprised 6.84% of the Eastern Division population but only 3.17% of those who identified race on returned questionnaires. In 2003, African-Americans comprised 6.96% of the Eastern Division population but only 3.17% of those identified race on returned questionnaires.

C. Steps Five and Six: Jury Impan-elment — Venire to Petit Jury Selection from 3% to Nil

Once questionnaires are returned and the available jury wheel is compiled, the federal Jury Administrator sends notices to appear to individuals on the available wheel randomly and at a pace consistent with the district’s need for jurors in any given month. Marking the fifth step in the jury selection process, the notice to appear directs potential jurors to report to the courthouse on a specified day for selection to a petit jury. Those jurors who appear at the courthouse on the date specified comprise the “jury venire.” Judges and parties then select citizens from the jury venire to serve on the trial jury (also referred to as the “petit jury”) the final step of the process.

Certain features of jury impanelment in capital cases are likely to aggravate any preexisting minority underrepre-sentation on the jury venire, an issue that will be dealt with more fully in a subsequent memorandum. For instance, evidence from previous federal capital trials in Massachusetts suggests that the process of death-qualifying 32 jurors depletes the already small number of African-American potential jurors. 33 Concerns about the impact of death-qualification may well be unique to this state; Massachusetts has no death penalty, and public opposition to it runs high. See, e.g., Frank Phillips, Support for Gay Marriage: Mass. Poll Finds Half In Favor, Boston Globe, April 8, 2003, at A1 (Massachusetts public opinion poll finds 41% opposition to capital punishment).

My November 4, 2004, decision ordering the impanelment of two juries, one to decide guilt and the other to decide punishment, was entered as a matter of case management to avoid the very complex death-qualification process. On May 12, 2005, the First Circuit Court of Appeals found jury bifurcation contrary to the plain language of the Federal Death Penalty Act. See United States v. Green, 407 F.3d 434, 2005 WL 1119791 (1st Cir.2005). 34 However, neither my decision nor the First Circuit’s opinion addressed potential constitutional issues stemming from death-qualification, particularly in light of an already underrepresented jury venire.

In the context of the findings of this decision — the decline in African-American representation from 20% to 7% (through the choice of a federal forum) and then from 7% to 3% (through the jury summonsing process) — the parties are ordered to brief whether death-qualification is likely to exclude minorities at such a high level as to raise renewed constitutional concerns not addressed by Supreme Court precedent; whether the Sixth Amendment applies at this stage of the proceeding at all; and if it does, whether there is a means less violative of defendants’ rights than the current approach to accomplish the government’s goals (namely, a bifurcated jury).

D. History of Minority Underrepre-sentation in Massachusetts and Related Federal Litigation

Scant minority representation on Massachusetts jury venires is not a new problem. Neither is the charge that the resident lists are not being updated annually, although it has never before been litigated as fully as in the instant case. While the problem has been recognized at the highest levels of government, all remedial efforts — judicial and extra-judicial- — that have been pursued over the years have apparently failed.

1. The 1993 Boston Litigation

In 1993, the OJC sued Boston over its refusal to comply with chapter 234A’s requirement to compile annual comprehensive resident lists for use by federal and state jury officials. See Jury Commissioner of the Commonwealth of Massachusetts v. Mayor of the City of Boston, et al., Suffolk Sup.Ct. No. 93-04718 (Mass. Dist. Ct. filed August 9, 1993). The litigation ended with the City’s agreement to employ “aggressive” and “diligent” efforts at compliance with the annual resident list requirement, including conducting door-to-door canvassing and follow-up mailings, as well as cross-checking public school enrollment, parking permits, and multiple-dwelling buildings. See id., Stipulation and Memorandum of Understanding, dated November 23,1994.

In the questionnaire defendants circulated pursuant to this litigation, see infra, Boston reported that it was fully complying with its obligations under the 1994 agreement. Defendants, however, suggest that precisely the opposite inference should be drawn from the dismal return rates for summonses mailed to Boston residents (62% for Boston versus 75% for neighboring Newton).

2. 1994 Supreme Judicial Court Gender and Race Bias Report

In 1994, the Massachusetts Supreme Judicial Court’s (“SJC”) Commission to Study Racial and Ethnic Bias in the Courts issued a report sounding themes very similar to those featured in the 1993 Boston litigation and the instant case. The Commission concluded that “[t]he failure of municipalities to comply with state law requirements to provide the [state] Office of Jury Commissioner with accurate, complete, and verified resident lists contributes to minority underrepresentation in jury pools.” Commission to Study Racial and Ethnic Bias in the Courts, Massachusetts Supreme Judicial Court, Equal Justice 55-68 (1994).

Specifically, the Commission noted inaccuracies in mailing addresses which “contribute [ ] to the large number of undeliverable summonses, particularly in poor neighborhoods.” Id. at 60. The study showed that Suffolk County had the highest rate of undeliverable summonses (24.8%) and the highest nonresponse rate in the state (22%). Id. at 63.

While both the Boston litigation and the SJC report found the underrepresentation of African-Americans in Massachusetts jury pools to be demonstrable, historically persistent and troubling, the situation has not improved over the intervening ten years.

3. Federal Litigation Between 1984 and 1999

In case after case over the past twenty years, the First Circuit concluded that defendants had not made out a prima facie case of a constitutional fair cross-section violation. In United States v. Hafen, 726 F.2d 21 (1st Cir.1984), the court rejected a claim that African-Americans were underrepresented based on a showing of a 2.02% difference between the percentage of African-Americans in the population and then-percentage on the jury wheel. In United States v. Pion, 25 F.3d 18 (1st Cir.1994), the court rejected a challenge to Hispanic representation on the jury venire, concluding that the defendants had not proven that the 3.04% difference between Hispanic representation in the population and the jury pool was attributable to anything but chance. Four years later, in United States v. Royal, 174 F.3d 1 (1st Cir.1999), the court found a 2.97% disparity between African-American representation in the population and on jury venires to be constitutionally indistinguishable from the disparity in Hafen, although this time the court noted that the statistics defendant presented were “disquieting” and indicative of “a situation leaving much to be desired.” Id. at 21.

E. Defendants’ Case

On November 1, 2004, at defendants’ direction, the federal Jury Administrator, James McAlear [“Jury Administrator” or “McAlear”], mailed questionnaires to the clerk of each of the 190 towns and cities in the Eastern Division. The questionnaire inquired as to the manner in which each town or city compiles its annual resident list. 35 While responses to the questionnaire were not given under oath, the government did not contest the accuracy of the facts asserted therein and did not object to admitting the completed questionnaires into evidence. Defendants also retained an expert: Andrew Beveridge, Professor of Sociology at Queens College and the Graduate Center of the City University of New York. Professor Beveridge is an expert in demography, particularly the statistical and quantitative analysis of U.S. Census data. He has testified as an expert in demographic and statistical analysis in nearly twenty court cases.

Professor Beveridge analyzed the available Eastern Division data for the years 2001 through 2003, 36 and the responses to the questionnaires. His analyses and conclusions were submitted to the Court in six affidavits, coupled with his testimony during the January 11 and January 26, 2005 hearings. McAlear, who is responsible for assisting the Clerk of the Court “in the performance of producing the master jury wheel” testified concerning his responsibilities during the same hearings. Jury Plan, at p. 1. The government did not enlist an expert and did not present any testimony.

Pursuant to Federal Rule of Evidence 706, the Court appointed Professor Jeffrey Abramson as an expert witness to assist the Court in analyzing the extensive data provided by defendants. See Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579, 595, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) (“Rule 706 allows the court at its discretion to procure the assistance of an expert of its own choosing”). Professor Abramson’s appointment became effective on February 18, 2005. See Order Re: The Court’s Intent to Appoint An Expert Witness, docket entry # 293, dated February 9, 2005.

Professor Abramson produced a report to the Court on April 21, 2005, concluding that the defendants had not made out either a prima facie constitutional violation or a violation of the JSSA. See Report on Defendants’ Challenge to the Racial Composition of Jury Pools in the Eastern Division of the United States District Court for the District of Massachusetts, docket entry # 329, dated April 21, 2005.

The parties were given until May 9, 2005, to file responses to Professor Abram-son’s report. See Order Re: Report Of The Court-Appointed Expert Witness, docket entry # 329, dated April 25, 2005. The government filed a response concurring in Professor Abramson’s conclusions. See Government’s Response to Court’s April 25, 2005 Order Regarding Report of the Courb-Appointed Expert Witness, docket entry #335, dated May 16, 2005. Defendants filed a response objecting to many of the conclusions reached in Professor Abramson’s report. See Response of Defendants Branden Morris and Darryl Green To The Report of Courb-Appointed Expert Jeffrey Abramson, docket entry # 337, dated May 17, 2005.

For purposes of the constitutional analysis, defendants contend that there is official misfeasance — haphazard or negligent preparation of the resident lists in the cities and towns with the highest African-American populations. 37 They could not provide direct evidence of official misfeasance in compiling resident lists. Only one town, New Bedford, blatantly admitted in its questionnaire to failing to update its resident list annually, as it is required to do. 38 The task of examining or cross-ex amining the other 189 city and town officials would have been a daunting one. Moreover, defendants did not have access to the 190 resident lists, nor would the lists have disclosed the race of each listed individual. Accordingly, defendants used the data from the only sources available to them—the master jury wheel, the available jury pools, and the grand jury pools, together with the rate of nonresponses and undeliverables. They argue, inter alia, that substantial differences between data and the resident census list data, coupled with high undeliverable and nonresponse rates, prove that the resident lists are woefully inaccurate.

Furthermore, defendants used zip code data to focus on precisely which cities and towns are suffering from the highest rates of undeliverables and nonresponses. They found these to be the cities and towns in the Eastern Division with the most African-American (and poor) residents. Thus, while virtually all of the 190 cities and towns in the Eastern Division claim to conduct an annual census, it appears that only the smaller, wealthier (and whiter) towns are conducting meaningful annual census counts. In contrast, the efforts of poorer towns are lackadaisical and inadequate.

The government blames the falloff in minority representation, and in particular the non-response rate, not on official misfeasance, but on demographic characteristics, like higher levels of transience among poorer populations, and private choices not to answer jury summonses. These characteristics, the government argues, cannot be redressed unless they result in a jury venire that fails to meet constitutional fair cross-section mínimums, or they are accompanied by statutory violations. The government does not believe that either condition is met here.

Notwithstanding the limitations of the data, one thing is clear: The data presented here represents the most comprehensive effort thus far to capture the causes of African-American underrepresentation in Eastern Division jury pools.

III. DISCUSSION

A. Fair Cross-Section Challenge

1. Fair Cross-Section Framework

The Sixth Amendment requires that juries are selected from pools representing a fair cross-section of the community. Duren v. Missouri, 489 U.S. 357, 363-64, 99 S.Ct. 664, 58 L.Ed.2d 579 (1979); see generally Nancy Gertner & Judith H. Mizner, The Law of Juries, §§ 2-11-2-19 (1997) [hereinafter Gertner & Mizner, The Law of Juries ]. Although petit juries need not mirror the exact demographic composition of the community, the process of selecting petit juries must give members of “cognizable” groups a fair opportunity to serve (i.e., they may not be systematically excluded from the pool). Taylor v. Louisiana, 419 U.S. 522, 538, 95 S.Ct. 692, 42 L.Ed.2d 690 (1975).

In Duren, the Court outlined the requirements for finding a prima facie violation of the Sixth Amendment, albeit in very general terms. Defendants must show:

(1) that the group alleged to be excluded is a “distinctive” group in the community [cognizable group prong]; (2) that the representation of this group in veni-res from which juries are selected is not fair and reasonable in relation to the number of such persons in the community [underrepresentation prong]; and (3) that this underrepresentation is due to systematic exclusion of the group in the jury-selection process [systematic exclusion prong].

439 U.S. at 364, 99 S.Ct. 664. Only if a prima facie violation is shown is the government required to justify its practices— to prove “that a significant state interest [is] manifestly and primarily advanced by those aspects of the jury-selection process ... that result in the disproportionate exclusion of a distinctive group.” Id. at 367-68, 99 S.Ct. 664.

There is no dispute that Duren’s first prong is satisfied here: African-Americans unquestionably constitute a cognizable group. See, e.g., Hafen, 726 F.2d at 23. The dispute here centers around Duren’s second and third prongs.

Lower courts have struggled to give content to Duren’s very general statements about underrepresentation and systematic exclusion. For the most part, the prima facie standards have been rigorously defined. The stringency of the prima facie standards is one explanation for the fate of Hafen, Royal and Pion — the First Circuit’s three hallmark (and unsuccessful) fair cross-section challenges — although there are recent indications that the Court could revisit its fair cross-section standards. See, e.g., Royal, 174 F.3d at 12-13 (statistics presented by the defendant were “disquieting” and “a situation leaving much to be desired”).

Put simply, Duren raises the following questions regarding the underrep-resentation prong: (1) By what statistical means shall the exclusion of members of a cognizable group be measured? (2) And, far more significantly, what level of exclusion of members of a cognizable group is tolerable for constitutional purposes? See Peter A. Detre, A Proposal for Measuring Underrepresentation In The Composition Of The Jury Wheel, 103 Yale L.J.1913 (1994) [hereinafter Detre, Measuring Un-derrepresentation ]. Precedent, including the First Circuit’s precedent, answers the second question with something of a contrivance, a normative determination of how much disparity is too much. As I describe below, there should not be a magic number. The answer should instead depend upon the context in which the underrepre-sentation is found, and upon the degree to which the goals embodied by the Sixth Amendment are or are not being achieved. This is fundamentally a qualitative, not a quantitative, inquiry.

With respect to question (1), the First Circuit has chosen to measure the under-representation prong using the absolute disparity test, which measures the difference between the cognizable group’s percentage in the relevant population against the group’s percentage on the jury wheel. 39 See Hafen, 726 F.2d at 23 (calculating “the difference between the percentage of eligible blacks in the population and the percentage of blacks on the master wheel”); see also Pion, 25 F.3d at 23 n. 5 (defining the absolute disparity standard as “the gross spread between the percentage of eligible Hispanics ... in the relevant population and the percentage of Hispanic representation on the Master Jury Wheel”). With respect to question (2), the court found that the absolute disparity of 2.98% in Royal was “not meaningfully distinguishable from the 2.02% absolute disparity accepted in Hafen." Royal, 174 F.3d at 10-11.

While the First Circuit surely did not adopt any given threshold talismanic figure, it has cited with approval United States v. Maskeny, 609 F.2d 183 (5th Cir.1980), which rejected an absolute disparity of 10%, and United States v. Clifford, 640 F.2d 150 (8th Cir.1981), which rejected an absolute disparity of 7.2%. See Hafen, 726 F.2d at 23-24.

With regard to Duren’s systematic exclusion prong, a court hearing a fair cross-section challenge must ask: (1) Is the un-derrepresentation caused by happenstance (which is not actionable), or is it caused by official action or inaction of some sort (which may be actionable)? (2) And if official misfeasance contributes somewhat to the disparity in representation, do the defendants bear the burden of showing precisely how much? There is little law on these issues because few cases have gotten beyond the underrepresentation prong.

One thing is, or should be, clear: Sixth Amendment analysis does not require proof that a cognizable group has been excluded because of discrimination, as in the case of an Equal Protection challenge under either the Fifth or Fourteenth Amendments. The distinction is important. An Equal Protection challenge concerns the process of selecting jurors, or the allegation that selection decisions were made with discriminatory intent. The Sixth Amendment, on the other hand, is concerned with impact, or the systematic exclusion of a cognizable group regardless of how benevolent the reasons. It looks to discriminatory effects, while the Equal Protection clause looks to discriminatory purposes. See Gertner & Mizner, The Law of Juries, at §§ 2-10-2-13. Even practices that are race-neutral but have a disparate impact on the representation of a cognizable class in the jury venire fit within the Sixth Amendment’s protections, while they would not be cognizable under the Equal Protection clause. 40

Defendants challenge First Circuit precedent on underrepresentation both with respect to the absolute disparity test and the percentages rejected by First Circuit precedent. In addition, they — along with Professor Abramson — suggest an alternative formulation, a hybrid test that melds the second and third prongs of Duren: If defendants can identify a mechanism by which a cognizable class is excluded — here, that city and town officials are not compiling annual resident lists that are even remotely accurate — -and if they can show that such misfeasance contributes to African-American underrepresentation in the jury pool, such a showing should suffice even if the absolute disparity is “only” 2 or 3%.

Below, I outline the law in this area (and its deficiencies), and then address defendants’ proof.

a. Second Prong: Underrepresentation

(1) By What Statistical Means Shall the Exclusion of Members of a Cognizable Group Be Measured?

In Hafen, the Court not only endorsed an absolute disparity analysis, 726 F.2d at 23-24, but also expressly rejected comparative disparity analysis, which measures whether there is a diminished likelihood that members of an underrepresent ed group will be called for jury service. 41 Id. (“Although we acknowledge the possibility that the comparative disparity calculation might be a useful supplement to the absolute disparity calculation in some circumstances, we do not believe that it necessarily produces a more accurate result where, as here, the group allegedly underrepresented forms a very small proportion of the total population.”). And in United States v. Royal, 174 F.3d 1 (1st Cir.1999), the Court refused to reconsider the absolute disparity model, even while conceding that the “case turn[ed]” on the choice of statistical methodology. 42 Id. at 5.

I am obliged to adopt the absolute disparity approach, although it is worthwhile to urge its reexamination. Absolute disparity analysis fails to capture the persistent underrepresentation in District of Massachusetts jury pools. 43 It clearly does not adequately address situations where, as in Massachusetts, the underrepresented group is a small percentage of the population. See United States v. Rogers, 73 F.3d 774, 776 (8th Cir.1996) (“Although utilizing the absolute disparity calculation may seem intuitive, its result understates the systematic representative defíciencies ...”); United States v. Jackman, 46 F.3d 1240, 1247 (2d Cir.1995).

Put simply, if each and every African-American in the Eastern Division were excluded from jury service, the absolute disparity would be “only” 6.96% — and within the range cited approvingly in Hafen. Such a result would legitimize the intuitively illegitimate — a jury trial without African-Americans for urban crimes that allegedly occurred in a county that is 20% African-American.

Defendants recognize that absolute disparity analysis must inform my review, but they urge me to supplement it with additional statistical models. 44 One such model is comparative disparity analysis. But here again, as noted above, the First Circuit has rejected comparative disparity analysis because it “distorts reality” where “a very small proportion of the population is black.” Hafen, 726 F.2d at 24 (“[T]he smaller the group is, the more the comparative disparity figure distorts the proportional representation.”); see also Royal, 174 F.3d at 7, 9 (citing cases from other circuits). 45 Of course, one could — and per haps should—draw precisely the opposite conclusion: Where a very small proportion of the population is African-American (largely because of the government’s choice of a federal forum), we have a heightened obligation to ensure that the highest numbers of that population will appear on our juries. See United States v. Levasseur, 704 F.Supp. 1158, 1162-63 (D.Mass.1989) (holding that “only a comparative disparity analysis will afford sufficient protection to defendants’ right to be tried by a fair cross-section of the community”). 46

Defendants also cite to statistical decision theory (“SDT”) and disparity of risk as potential supplements to absolute disparity of risk. 47 Using binomial distribution, SDT calculates the probability that an observed underrepresentation occurred by chance. 48 In other words, SDT is a test of whether underrepresentation could be the result of a random process rather than a systematic failure; the smaller the probability produced by the SDT calculation, the less likely it is that the observed underrep-

resentation occurred by chance. Defendants assert that the average SDT for the relevant data from 2001 through 2003 is 1 in over 507 million.

Also using binomial distribution, disparity of risk describes the increase in a defendant’s chance of drawing an underre-presentative petit jury as a result of an underrepresentative jury pool. It measures the likelihood of having at least one African-American juror in a given twelve-member jury. Defendants calculate that, with a fully representative wheel for the years 2001 through 2003, approximately 58% of all twelve member juries in the Eastern District would have at least one African-American juror. But with the alleged underrepresentation of African-Americans in the available jury wheel from 2001 through 2003, defendants assert that an average of only 29.48% of juries would include at least one African-American juror. Thus, defendants argue that the likelihood of underrepresentation is both more prevalent and more entrenched than absolute disparity alone indicates. 49

The First Circuit has never addressed either SDT or disparity of risk. See Royal, 174 F.3d at 7 n. 3 (“Our case involves only these two methodologies [absolute disparity and comparative disparity] — the only two discussed by the parties — and our endorsement of one should not be taken as a statement that it is the best of all possible methodologies”). While use of SDT is not without precedent in other circuits, see e.g. United States v. Jackman, 46 F.3d 1240, 1247 n. 5 (utilizing a derivative of SDT), both SDT and disparity of risk have faced significant criticism. 50

I urge the First Circuit to supplement the absolute disparity test with other statistical measures or the hybrid approach described below. No methodology has been mandated by the Supreme Court. Nor is there anything about absolute disparity that logically favors it over a more complex analysis of underrepresentation. More significantly, there is every reason to reconsider the approach at this time. To suggest that the relentless pattern of criminal defendants confronting white, or largely white, juries does not deserve further constitutional scrutiny is a troubling act of judicial complicity. In 1984, in Hafen, the pattern may not have been clear. Today, two decades later, it is.

(2) How Much Exclusion of Members of a Cognizable Group Is Significant for Constitutional Purposes?

The First Circuit has not adopted a threshold for answering the question of how much exclusion of members of a cognizable group is significant for constitutional purposes. However, it has told us that 2.02% absolute disparity in Hafen, 726 F.2d at 23, and 2.97% in Royal, 174 F.3d at 10-11, were not enough. 51

A fair cross-section analysis should not be about picking a number out of context; it should be about how much exclusion of a cognizable group the Constitution should tolerate. Plainly, an entire group does not have to be eliminated from the jury pool before constitutional alarms are raised. But how much underrepresentation is too much? Perhaps we should take into account the fact that the choice of forum— the Executive’s choice — has already altered the decisionmaker, from a jury pool that is 20% African-American to one that is 7%. Perhaps we should care that, whatever the numbers, the vast majority of Eastern District juries will not have a single African-American member. 52

The function of a prima facie case is to set the minimum threshold of evidence a moving party must offer before the court scrutinizes the practice further. In the context of this case, the bar should be set lower. Where African-American representation in the pool is effectively halved, and where there is evidence of systematic defects, the court should look more closely and the government should bear the burden of justification.

b. Third Prong: Systematic Exclusion

(1) Is the Disparity Found under the Second Prong Caused by Happenstance (Which Is Not Actionable) or Is it Caused by Official Action or Inaction of Some Sort (Which May Be Actionable)?

The facts of Taylor and Duren illustrate what sorts of systematic defects qualify as potential violations of the Sixth Amendment. In Duren, administrators determined that it would be more convenient to give women an automatic exemption once they requested not to serve, since they were likely to claim exemptions based on child-rearing obligations. 439 U.S. 357, 99 S.Ct. 664, 58 L.Ed.2d 579 (1979). And in Taylor, women were required to specially register for jury duty. 419 U.S. 522, 95 S.Ct. 692, 42 L.Ed.2d 690 (1975). It may have been true — especially at the time of these cases — that, because women were more likely to be home with children, “hardship” disqualifications that were neutral on their faces would have led to their exclusion anyway. But plainly the government cannot put its fingers on the scale. The Supreme Court in both cases found that the government action cannot exacerbate a pattern of exclusion, even one enshrined in the culture. 53

In United States v. Pion, the First Circuit found that the defendant had not identified “a systemic defect” or “an operational deficiency in the Jury Plan which would account for the alleged underrepresentation.” 25 F.3d at 23. Because names for the master jury wheel were drawn from resident lists — what the court deemed the broadest data available — there was no reasonable inference that the small number of Hispanics in the pool was attributable to “anything other than the randomness of the draw.” Id. at 23-24. In contrast, defendants here do identify an operational deficiency, namely, the inaccuracy of resident lists that form the source of the jury pool, due to official negligence (or worse). They claim that the resident list approach is an improvement over the use of voter lists only to the extent that it is executed properly.

Still, the government interprets the evidence to suggest that the cities and towns complied with the law, and carried out all statutory mandates, but still could not capture a more transient African-American community or persuade more African-Americans to return questionnaires. Since there is no constitutional requirement to take affirmative steps to ensure a representative jury, according to the government, there is no violation.

The parties’ different perspectives beg the question: What if the data suggests that both the government and the defendants are right, that the problem is attributable to some degree of official misfeasance, and some degree of demographic factors? Is that sufficient?

(2) Even if Official Misfeasance Contributes Somewhat to the Disparity in Representation, Do the Defendants Have to Show Precisely How Much of the Disparity Is Attributable to Such Factors?

Defendants claim that they do not have to prove that systematic defects are responsible for the totality of the problem. In other words, while there may be demographic reasons for the underrepresentation of African-Americans, where official action or inaction has exacerbated the problem, they have made out a prima facie case. The government argues, and the court-appointed expert agrees, that defendants have to prove the precise extent to which official negligence or misfeasance is responsible for the disparity to meet constitutional standards.

This debate is similar to the one in employment law about “mixed motive” claims: Does a plaintiff have to show that a given adverse employment decision was entirely caused by a defendant’s discriminatory animus? Or is it enough to show that discriminatory animus played a part in the final decision? In this context, if the goal is a fully representative jury, it should be enough that official misfeasance played a part in diminishing African-American representation, even if we cannot quantify that role, much less effect a perfect system because there will always be some people who will not respond to questionnaires or who will frequently change residence. Indeed, as Professor Abramson says, a more exact test would be well nigh impossible for defendants to meet. Abramson, Report, at 4 n. 2 (“[T]he data on the record, though detailed and probably the best that is available, cannot with mathematical certainty tell the Court exactly what percentage of a given town or zip code’s undercount is African American”).

First Circuit case law has not addressed this issue. However, given the rigors of the First Circuit’s approach on other fronts, I can only assume that they would adopt Professor Abramson’s approach.

c. Hybrid Approach

The case law suggests that the representativeness and the systematic exclusion prongs are independent of one another: Traditionally, if the absolute disparity is not high enough, a court may not even address the mechanism of exclusion. But there is a hybrid approach, as implied by the analysis above and suggested by Professor Abramson. He states:

[A]s a matter of law, there is some interplay between the “substantial” and “systematic” prongs of the Duren test... Thus, the more clear it is that the underrepresentation of a cognizable group is caused by the kind of official misfeasance that defendants allege here, the less tolerance there ought to be for loss of fair representation for that group.

Abramson, Report, at 14.

In effect, the hybrid approach views the second and third prongs of the Duren test as relating symbiotieally with one another. If defendants are able to identify practices that serve systematically to exclude African-Americans from the jury pools of the Eastern Division, courts should view defendants’ absolute disparity data more favorably than they would in the absence of identified exclusionary practices. This approach finds support in the case law and is an avenue of analysis that has not been foreclosed by the First Circuit. See, e.g., United States v. Rioux, 930 F.Supp. 1558, 1566 (D.Conn.1995) (“[T]he second and third prongs of the Duren test, unfair representation and systematic exclusion, are intertwined inextricably”); Commonwealth v. Arriaga, 438 Mass. 556, 566, 781 N.E.2d 1253 (2003) (“Evidence of a disparity smaller than 10% can support a conclusion of unconstitutional underrepresentation of smaller minority groups, especially when coupled with persuasive evidence of systematic exclusion”) (emphasis added); see also United States v. Biaggi, 909 F.2d 662, 679 (2d Cir.1990), cert. denied, 499 U.S. 904, 111 S.Ct. 1102, 113 L.Ed.2d 213 (1991).

2. Defendants’ Case

a. First Prong: Distinctiveness

[17] Because African-Americans are a distinctive group, defendants have unquestionably satisfied the first prong of the Duren test. See, e.g., Hafen, 726 F.2d at 23.

b. Second Prong: Underrepresentation

Defendants have calculated an absolute disparity of 3.67% for 2001, 3.58% for 2002, and 3.79% for 2003 in African-American representation in the Eastern Division. Taken together, these figures amount to an average absolute disparity of 3.66% for 2001 through 2003. 54 Defendants concede that such numbers would not ordinarily be sufficient to meet the second prong of the Duren test, given the First Circuit’s position in Hafen and Royal.

As Professor Abramson noted, however, the question of whether the Sixth Amendment (or the JSSA) is offended by this absolute disparity “cannot be answered in a vacuum.” Abramson, Report, at p. 23. Consider the following:

First, the decline from a 6.96% share of the adult population to a 3.17% share of names on the available jury wheel “means that less than half the numbers of African-Americans that random selection should have placed on the available jury wheel are in fact present there.” Abramson, Report, at p. 23.

Second, it “means that the defendants in this case suffer a statistically significant decline in the probability that their jury venire and their jury selected from the venire will have the number of African-Americans on it that random selection from an initially representative pool would be expected to produce.” Id.

Thus, as Professor Abramson states:

These problems are serious enough that, rather than simply saying formulaically that absolute disparities on the order of 3.79% (i.e. 6.79 minus 3.17) automatically pass muster under the Constitution and the Act, the Court should consider who or what is responsible for a less than desirable result and what difference it might legally make if it were clear that official state action, and not mere private choices, lay behind even a 3.79% absolute disparity.

Abramson, Report, at p. 24. Using a hybrid approach, I turn to the question of whether there is systematic exclusion and to what degree it is responsible for the disparity between ideal and actual African-American representation in the jury pool.

c. Third Prong: Systematic Exclusion

What is novel about defendants’ fair cross-section challenge is their evidence of systematic defects in the way the source lists (resident lists) are prepared. Defendants present persuasive quantitative evidence that, as a result of these defects, the resident lists are not functioning as the Jury Plan assumed they would. In addition, the questionnaires that were sent to every city and town bolster the conclusion that the cities and towns vary in the way they update their annual lists — if they prepare annual lists at all. 55 As a result, instead of capturing an accurate cross-section of the community, the resident lists are skewed because officials in some towns (white, affluent, and suburban) are doing a better job of continually updating and improving their lists than are officials in other towns (African-American, poor, and urban).

Defendants document three phenomena that cause African-American underrepre-sentation: 1) overall, the resident lists un-dercount African-Americans from the outset; 2) many of the resident lists are not improved and updated annually, as required by state law, resulting in a disproportionately high rate of undeliverables (and nonresponses) in heavily African-American, poor, and urban communities, where geographic mobility is high; 3) the decision of some cities — notably Boston'— to purge inactive voters from their resident lists, while other cities and towns include inactive voters on their resident lists, aggravates undeliverable and nonres-ponse rates.

Each of these allegations is addressed in turn.

(1) Shortcomings of The Resident Lists — Undercounting and Over-counting

To illustrate how overcounting and un-dercounting of residents have racially disparate effects, defendants created a subset of 35 towns (out of 190) in the Eastern Division. Twenty-one of those towns represent all Eastern Division towns that have fewer than 500 African-American residents of voting-age, less than 3% of their residents living in poverty, and a population of more than 10,000 (“the 21 towns”). The remaining 14 towns are those with more than 2,000 African-American residents of voting-age, more than 4% of residents living in poverty, and a population of more than 10,000 (“the 14 towns”). 56 The combined resident lists for the 21 towns overrepresented their Census-expected population (103.48% of the Census estimate), while the combined resident lists for the 14-town set yielded only 85.3% of their Censures estimated population.

The data is even more telling at the zip code level because one can infer more from smaller, relatively more homogenous units. 57 For example, Dorchester (02121), where African-Americans constitute 83% of the population, is undercounted and thus underrepresented on the jury wheel by 22%.

Both the government and Professor Abramson argue that undercounting and overcounting is not necessarily a primary cause of African-American underrepresen-tation. They explain that defendants’ data falls short of proving that African-American residents are being undercounted more than the other residents of the 21 towns. There is, according to Professor Abramson, a problem of “ecological inference” — seeking to draw inferences about a fraction of the whole using aggregate data about that whole. 58 To draw definite conclusions based on aggregate data, there would need to be perfect racial residential segregation, which there is not. See G. Thomas Munsterman and Daniel J. Hall, Jury Management Study: Kent County, Michigan (2003).

The government argues that even zip code data — the smallest fraction about which reliable geographic data is compiled — does not sufficiently disaggregate residential demographics. While such data may be telling for homogeneous zip codes, like Dorchester, it does not say much about the effects of over and under-counting in the dozens of zip codes that are racially integrated, like Jamaica Plain (02130), where African-Americans constitute 16% of the population.

The issue, as I have noted, turns on the rigor of the constitutional test. If defendants can pinpoint where the undercount-ing is occurring and can even show it substantially affects African-Americans, as they have, but are required to show precisely how much (i.e., the extent to which African-Americans are disadvantaged by undercounting rather than other groups), then this challenge will fail. And if they have garnered the best available data, better than any other defendant to date, then every comparable challenge will also fail. Again, if that is the case, perhaps the constitutional test should be reexamined.

(2) Shortcomings of Summonsing— Demographics, Logistics and Non-responses

Defendants demonstrate that the greatest drop in the proportional representation of African-Americans in Eastern Division jury pools occurs during the process when individuals are summonsed from the master jury wheel. In other words, the group that arrives at the federal courthouse for impanelment is far whiter, in terms of percentages, than the master jury wheel.

Boston, with its large percentage of African-Americans (25.3%) is illustrative: Boston’s share of the names on the master jury wheel is 11.1% less than it would be if it were proportional to Boston’s Census-estimated share of the Eastern Division’s voting-age population. After the summonsing process, Boston’s share of the jury venire drops to 28.79% less than it would be if the pool perfectly represented the Eastern Division’s voting-age population. The four zip codes in Boston with the highest absolute number of African-American residents begin with an under-representation on the master jury wheel ranging from 14.83% to 22.57%. But the initial underrepresentation in these zip codes swells to a range of 31.90% to 41.84% by the time juror questionnaires are returned and the available jury wheel is compiled. Patently, the process of summonsing jurors exacerbates whatever un-derrepresentation Boston already experi- enees at the resident list and master jury wheel stages.

Again, the question is why this is so, and how far defendants must go to prove these shortcomings are to be deemed “systematic.” Is it enough to show that some of the underrepresentation is attributable to summonsing from outdated and inaccurate city and town lists, producing disparate rates of undeliverable mail in the more racially diverse areas? Or is a finding of a “systematic” exclusion defeated if factors that are beyond the federal court’s province are also to blame, like high levels of transience among poorer populations and individual choices to ignore jury summonses? Should it be enough to show that factors, like transience and individual choice, are exacerbated by official action or inaction — as in Duren and Taylor?

(a) Undeliverables

With regard to undeliverable summonses, the combined data from 2001 through 2003 shows that 12.4% of all jury summonses were returned by the post office marked “undeliverable.” In contrast, 15% of all summonses mailed to addresses in Suffolk County, where a large portion of the Eastern Division’s African-American population resides, were returned as undeliverable, and 15.2% of all summonses mailed to addresses in Boston proved undeliverable. For the 21 towns (with smaller African-American populations), the undeliverable rate was 5.8%, while the undeliverable rate in the 14 towns (with larger African-American populations) was 18.4%.

Accordingly, defendants argue that it is more likely that the addresses of African-Americans will be incorrect or outright missing from the resident lists, thereby systematically leading to higher rates of undeliverable summonses in cities and towns with high concentrations of African-American residents. In essence, names on the master jury wheel that do not have a corresponding address or have an incorrect corresponding address might as well not be there at all.

To be sure, demographic factors also play a role. Towns with higher poverty rates and larger concentrations of African-American residents tend also to have higher-than-average rates of mobility. For the 14 towns in the 35-town subset, the rate at which persons move within two years is 8.7% higher than the moving rate in the 21 towns. 59 In other words, as the government argues, even if officials did everything they could to update their resident lists, undeliverable rates would still be positively correlated with poverty rates and the size of the African-American population.

(b) Nonrespondents

Nonrespondents represent those persons summonsed for jury duty who fail to respond even though their summonses are not returned to the post office marked as “undeliverable.” Nonrespondents may be the product of wrong addresses, or individuals who chose not to answer the summons at all. This data is the most difficult to interpret.

In the Eastern Division, from 2001 through 2003, an average of 12.2% of pre sumably successfully delivered summonses produced no response. In contrast, the nonresponse rate for Suffolk County from 2001 through 2003 was 22.43%, and the nonresponse rate in Boston for that time period was 23.06%. For defendants’ subset of 35 towns, the 14 towns (heavily African-American) had a nonresponse rate of 16.9%, while the 21 towns (heavily white) had an average nonresponse rate of 7.6%. At the neighborhood level, of the 32 zip codes in Suffolk County, 16 had nonres-ponse rates in excess of 20% (14 of these zip codes are in Boston).

If only some of these summonses were actually delivered, then the resident lists are even worse than the undeliverable rates suggest. The government argues, however, that if we assume that most of these summonses were delivered, then high nonresponse rates demonstrate that disparate rates of response among racial (and perhaps economic) groups skew the racial composition of the available jury pool further. For this, the government suggests, systematic exclusion cannot be blamed. 60

Again, the issue is the legal standard: It may not ever be possible to disentangle choice and transiency from bad lists for constitutional purposes. In any event, as described below, in connection with the JSSA, even high nonresponse rates do not absolve federal officials from taking steps to improve or otherwise compensate for African-American underrepresentation on jury pools, particularly where there are practicable steps that can be taken.

(3) Inactive Voter Lists

In the course of litigating their challenge, defendants learned that the Eastern Division’s cities and towns submit their resident lists to the State Central Voter Registry. The Election Division then forwards two electronic files for each city and town to the OJC. One file is the city or town’s list of all residents and active voters, and the other is a list of inactive voters. The OJC leaves it to each city or town to decide whether the names in the inactive voter file should be merged with the names of residents and active voters to form the final resident list. 61

In 2003,152 cities and towns (out of 190) requested or had standing orders to include names of inactive voters on their resident lists. Boston was among the minority of towns not requesting inclusion, and it in fact purged 99,716 inactive voters from its 2003 resident list. The result of this haphazard system, according to defendants, is that some towns inflate their resident count by including inactive voters who have moved, while other towns, like Boston, under count their residents by purging all inactive voters, many of whom may still reside in the city.

This phenomenon is worrisome insofar as it shows, at a minimum, that the very definition of “resident list” varies from town to town in ways the District intended to prevent in enacting the Jury Plan. But there is little basis for concluding that the choice to purge inactive voters amounts to systematic exclusion causing the underrep-resentation of African-Americans on the jury venire. If some towns are listing phantom residents by including inactive voters, then one would expect this to be revealed in higher than average rates of undeliverable mail and nonrespondents. The record suggests the opposite.

3. Conclusion: Defendants Have Not Made Out A Constitutional Violation

Defendants’ data suffers when judged by the insurmountable rigors of the existing Sixth Amendment framework. Even if defendants met the absolute disparity threshold — which would be nearly impossible — they would have had to prove not merely that inaccurate lists contribute to a degree to the underrepresentation of minorities, but the precise degree to which they are responsible, i.e., that inaccurate lists (as opposed to transience or personal choice) are the main culprit. That burden is far, far, too high; particularly when a prima facie case is involved.

It should be enough, as it was in Duren and Taylor, to demonstrate that there are clear procedural defects in the way the resident lists are compiled, and that these defects disproportionately affect a cognizable group, leading to that group’s significant underrepresentation in jury pools. But it is not, under current First Circuit law, and thus defendants have failed to make out a prima facie case of a constitutional violation.

The JSSA’s “substantial failure to comply” provision, however, is a different matter.

B. Statutory Challenge

1. The JSSA’s Proportionality Requirement

The JSSA defines two sorts of obligations — a requirement of proportionality, and a requirement of substantial compliance with the provisions of the Act. I address the former first.

The JSSA requires that every district’s jury plan “ensures that each county, parish, or similar political subdivision within the district or division is substantially proportionally represented in the master jury wheel for that judicial district [or] division.” 62 28 U.S.C. § 1863(b)(3). The Act further specifies that proportional representation can be measured by using “either the number of actual voters at the last general election in each county, parish, or similar political subdivision, or the number of registered voters if registration of voters is uniformly required throughout the district or division.” Id.

The question raised by this case is how to define “proportionality”: Is it simply the requirement of random selection? If the source lists are voter lists, is it enough for officials to make certain that the master wheel reflects random selection from those lists? If the source lists are resident lists, does proportionality simply require random selection from those lists? Or does proportionality mean something more—that the court is to measure the efficacy of the plan in terms of whether it accurately captures the actual population as reflected by Census data, rather than source lists?

At the heart of the debate is the significance of the 1992 Amendment to the Act, which provided that the District of Massachusetts may use numbered resident lists, rather than voter lists, as its source of names for the federal master jury wheel. 28 U.S.C. § 1863(b)(2). The government argues that the amendment merely had the effect of changing the benchmark for measuring proportionality from voter lists to resident lists. It reaches this result through parallel reasoning: The Act directs most federal judicial districts to use voter lists as a source of juror names, thereby making the test of proportionality in those districts whether their master wheels preserve the county (or other relevant political divisions) proportions as they exist on the voter lists. Thus, when the Act authorized the District of Massachusetts to replace voter lists with numbered resident lists, it implicitly made resident lists the measure of proportionality.

Defendants argue that, because the Act states that proportionality “may” be measured against benchmarks such as the previous year’s voter lists or resident lists, the language of the Act is discretionary. They urge the Court to employ this discretion, in conjunction with its supervisory authority, to determine that the better test of proportionality compares county proportions on the master jury wheel to county proportions as measured by Census counts. If the addresses on the master wheel are inaccurate, they argue, proportionality is irrelevant; the county population figures are chimerical.

The government’s interpretation makes sense in light of the plain language of the Act, although it clearly guts the proportionality requirement of any real meaning. In adopting the proportionality requirement, it was Congress’ intent that the “initial source lists ... accurately mirror community makeup.” House Report, 1968 U.S.C.C.A.N. 1792, at 1794. If the initial source lists are missing large portions of the population and consequently fail to “mirror community makeup” at the outset, as is the case here, they cannot be used to measure proportionality in a meaningful way. Unless proportionality is measured with respect to the percentage that each county actually represents of the Eastern Division’s entire adult population, the policy of the Act is defeated. In effect, under the government’s definition, proportionality would require no more than sound arithmetic. 63

[19] In resolving defendants’ proportionality challenge, the Court need not decide this issue of first impression—whether the JSSA grants the district court the discretion to employ something more meaningful than resident and voter lists for measuring proportionality—because the master jury wheel percentages deviate only a small amount from Census counts when county representation is considered.

According to defendants’ own data for 2003, the variances between county percentages according to Census estimates and county percentages as they actually existed on the master jury wheel appear to be less than one percent.

Other courts have found that relatively small departures from proportionality as measured against Census figures are merely technical, and thus not actionable. See United States v. Bailey, 76 F.3d 320, 322 (10th Cir.1996); United States v. Rosenthal, 482 F.Supp. 867, 871 (D.Ga.1979).

Defendants argue, however, that the important units for determining proportionality in this case are not counties, but smaller localities within Eastern Division counties, like cities, towns, and even zip codes. In 2003, for example, the City of Boston submitted 31,003 fewer names to the OJC than it should have, according to U.S. Census estimates, yielding a negative 6.75% deviation. For the same year, the City of Chelsea submitted 10,932 fewer names than their estimated Census population, yielding an alarmingly negative 44.11% deviation.

In United States v. Foxworth, 599 F.2d 1 (1st Cir.1979), the First Circuit rejected a locality-level proportionality challenge even where there was evidence that certain towns, not clustered in any one county, had been omitted from the jury wheel wholesale. The defendant in Foxworth argued that “certain cities and towns in the Eastern Division [...] were not represented in the Master Jury Wheel.” Id. at 2. In denying the defendant’s challenge, the court reiterated that § 1863(b)(3)’s requirement of “substantial ] proportionality]” applies only to county-level representation on the jury wheel:

Defendant has made no showing, un-sworn or otherwise, that any county or parish was not properly represented in the jury pool. It is not suggested that a city or town is a political subdivision similar to a county or parish. Nor has defendant shown that any lack of representation of the excluded cities and towns significantly affected the proportional representation of the counties within which such cities and towns are situated.

Id. at 4.

Without more profound disparities at the county level, the Act’s proportionality requirement cannot be used to redress the falloff that defendants have correctly identified.

2: “Substantial Failure to Comply” with the JSSA

Under 28 U.S.C. § 1867(d), if the court determines that there has been a “substantial failure” to comply with the provisions of the Act, “the court shall stay the proceedings pending the selection of a petit jury in conformity with this title.” The “substantial failure” to comply provision is broadly drafted, referring to all aspects of the JSSA.

Defendants argue that they have established a “substantial violation” of the Act, even if they have not established a constitutional violation, based on the failure of federal officials to supplement the available jury list despite its demonstrable deficiencies. They point to the JSSA’s requirement that the district’s jury plan “prescribe some other source or sources of names” in addition to the usual source list “where necessary to foster the policy and protect the rights secured by” the fair cross-section guarantee. 64 28 U.S.C. § 1863(b)(2); 28 U.S.C. § 1861.

Morever, even though the Jury Plan of the District of Massachusetts does not explicitly mention the statutory duty to supplement from other sources, it includes two provisions that reflect comparable, if not identical, concerns. Section 11(d) permits the Court to direct the Clerk to draw a supplemental array from the master wheel, to be added to the regular array “as necessary” when additional names are needed “because of excused or increased jury requirements.” Section 7(a) gives the Clerk, in consultation with the Chief Judge, the option of implementing a data processing system that will, inter alia, ensure that, in “[t]he selection of names from the source list and the master wheel ... the mathematical odds of any single name being picked are substantially equal.” 65

Plainly, the “mathematical odds” that any given name will be picked are not “substantially equal” across the region, as the Plan requires. A citizen’s chances of getting on a federal jury are far better in Needham and Sudbury, which update their lists through multiple mailings, telephone calls, and even door-to-door visits, than they are in Lynn and Brockton, where there is no follow-up, or in New Bedford, where there has not been an annual count since 1999. Nor has the random selection of names from resident lists, rather than voting lists, been sufficient to guarantee a fair cross-section of the community. If the Lynn, New Bedford and Brockton resident lists are inaccurate, it does not matter how perfectly random the procedures used by the Court are to cull names from resident lists; the names on the list are chimerical. Finally, these deficiencies are hardly insubstantial; they result in high rates of undeliverables and nonresponses to jury summonses, particularly in communities with large African-American populations.

Defendants propose several sorts of supplementation measures in this case. They include short-term measures that can be easily addressed in the case at bar, within the four corners of the Jury Plan and this Court’s supervisory authority, and longer-term measures that should be addressed by the Clerk in conjunction with the Chief Judge, and the reviewing panel of the Jury Plan described in 28 U.S.C. § 1863(a). 66

The short term proposals are:

a) For all summonses returned to the Court as “undeliverable,” the same num ber of new summonses should be mailed to residents who live in the same zip code area as the undeliverable summonses targeted;

b) For all summonses returned to the Court as “nonresponses,” the Jury Administrator should take steps to determine if the nonresponders in fact live at the listed addresses or if these are also “undelivera-bles” (i.e., sent to inaccurate addresses); 67

c) As an alternative to (b), additional summonses should be sent in numbers equal to the number of nonresponses in a given zip code area without determining whether the nonresponses are in fact un-deliverables.

In addition, defendants propose that the federal authorities inform the OJC which summonses have been returned as undeliverable, in the hopes that the source lists can be updated and inaccurate addressees expunged.

The longer-term proposals are:

a) The weighted mailing approach: The jury staff will compare the census percentages per each zip code with the percentages on the master wheel; they will then send out additional summonses to certain zip codes to proportionally reduce the chances that someone in an overrepresented zip code will receive one; 68

b) The “under yield” approach: The jury staff will adjust the numbers of sum- raonses based not on Census figures but on historical response rates; the response rates will be calculated based on the difference between the number of mailed summonses and the number of returned questionnaires.

I will order a) and c) of the short-term proposals in connection with jury selection in the instant case, as well as the proposal for informing the OJC of flawed addresses. Unless the jury staff reports to me otherwise, I will assume that these measures can be accomplished without delaying jury selection, scheduled to begin on September 19. While I am nominally ordering a stay under the statute, “pending selection of a petit jury in conformity with this title,” 28 U.S.C. § 1867(d), I have every confidence that the statutory purposes can be achieved without a continuance of this trial.

I will submit the longer range issues to the Clerk, the Chief Judge, and the statutory reviewing panel for further study. In effect, these alternatives are the institutional equivalents of the ad hoc adjustments I am making in the case before me. They involve more than simply counting the numbers of undeliverables and nonres-ponses in a given case. The long-term proposals require that jury staff review historical data or differentials between the current master jury wheel and Census counts, and adjust the summonses accordingly for all cases in this jurisdiction. Since they more directly implicate the terms of the Jury Plan, the “method and manner of random selection” under the Plan, they should be considered by other decisionmakers.

Finally, I will also urge our Court to work with state authorities to address these problems at their source — the unfunded mandate of cities and towns to prepare accurate residential lists.

In the final analysis, Professor Abram-son sums up the issues best:

However the Court rules on the motion before it, it is certainly worth the attention of the District Court as a whole as to whether public confidence in the integrity of the jury system is undermined by a jury selection process that is not truly random but makes chances of being called vary according to the diligence of one’s town in listing all residents.

Abramson, Report, at 26.

In the subsequent sections, I describe in detail what is a “substantial” violation of the Act, what the duty to supplement the source list comprises, whether the duty has been violated in this case, and the remedy under the JSSA. Finally, I address each of the objections raised by the government point by point.

a. What Amounts to a “Substantial” Violation of the Act?

A substantial failure under the Act is not a violation as to which the defendant must prove prejudice, on the one hand, 69 nor is it merely a technical violation of the Act’s strictures, on the other hand. 70 It is something in-between, which violates the letter and spirit of the JSSA. The statute plainly leaves it up to the courts to determine the meaning of “substantial failure to comply.” See 28 U.S.C. § 1867(d); United States v. Calabrese, 942 F.2d 218, 227 (3rd Cir.1991) (“Congress left the content of [substantial] largely up to the courts,” 942 F.2d at 227, citing House Report, 1968 U.S.C.C.A.N. at 1794 (“Your committee would leave the definition of ‘substantial’ to the process of judicial decision”)).

b. Section 1863(b) (2)’s Duty to Supplement

(1) The Statutory Language

The JSSA requires that district court jury plans supplement district source lists “where necessary to foster the policy and protect the rights” of litigants to a jury that reflects a fair cross-section of their community. 28 U.S.C. § 1863(b)(2). The language is that of an affirmative obligation to ensure selection from a fair cross-section of the community. There is no requirement to prove causation or to identify the mechanism by which the underrepresentation occurs. There is no inquiry as to whether the underrepre-sentation was created by discrimination or by the choice of the potential jurors (in not registering to vote, for example, in a jurisdiction where voting lists are used). Rather, the obligation to supplement is triggered by any set of facts which undermines the “policy” of the JSSA and the “rights secured by sections § 1861 and § 1862.” 28 U.S.C. § 1863(b)(2). Those rights include freedom from discrimination (§ 1862), as well as cross-sectional jury selection (§ 1861).

(2) Legislative History

The legislative history of the JSSA is consistent with an affirmative obligation to ensure a fair cross-section. 71 Before Congress enacted the JSSA in 1968, most jurisdictions used the “key-man system,” which relied on “key men” in the community to supply the jury commissioner with names of potential jurors. While the system had been criticized as capable of abuse, the Supreme Court had not found that the system qua system violated the Sixth Amendment. See Akins v. Texas, 325 U.S. 398, 65 S.Ct. 1276, 89 L.Ed. 1692 (1945); Smith v. Texas, 311 U.S. 128, 61 S.Ct. 164, 85 L.Ed. 84 (1940); see also Carter v. Jury Comm’n, 396 U.S. 320, 90 S.Ct. 518, 24 L.Ed.2d 549 (1970).

The Fifth Circuit in Rabinowitz v. United States, 366 F.2d 34, 56-57 (5th Cir.1966), however, took a different approach. It suggested that, even absent discriminatory intent on the part of the “key men” who did the jury selecting, the system had to be abandoned. The court noted that “the Constitution and laws of the United States place an affirmative duty on the court clerk and the jury commissioner to develop and use a system that will probably result in a fair cross-section of the community being placed on the jury rolls.” Rabinowitz, 366 F.2d at 57 (emphasis added).

Congress passed the JSSA in response to Rabinoioitz, doing via statute what the Supreme Court had chosen not to do through its constitutional analysis. See S.Rep. No. 891, 90th Cong., 1st Sess. 10-11 (1968) [hereinafter “S. Rep.”]; H.R.Rep. No. 1076, 90th Cong., 1st Sess. 4 (1968) [hereinafter “H.R. Rep.”]. The legislative history is extraordinary, reflecting several important themes, which have perhaps been forgotten in the administration of the JSSA:

First, Congress decried the underrepre-sentation of minorities in the “key-man” system, whatever the cause, whether intentional discrimination, or the natural, even well-intentioned, tendency of the key men to draw upon their limited circle of acquaintances. H.R. Rep. at 4; S. Rep. at 10. The statute’s goal was broad and remedial: “The defect that calls for congressional action is that the representational goal of jury selection is impaired when the methods used are haphazard or less than adequate to ensure fair selection from a fair sample.” S. Rep. at 10 (emphasis added).

Second, even though Congress chose to use voter registration lists as the default source list, it recognized their inadequacies. It left it to the courts to define when a particular voter list is so underrepresen-tative that it requires supplementation, as in communities in which a substantial percentage of the population has not registered to vote. S. Rep. at 16-17, 25.

Third, and relatedly, Congress recognized that the fact that a citizen chose not to register to vote did not necessarily mean it was appropriate to disqualify him or her from jury service. The issue was not that potential juror’s individual choice. Rather, it was the right of the defendant to a jury drawn from a fair cross-section of the community; a goal that would be undermined if juries consisted exclusively of those “who have manifested their civic interest by registering to vote.” 72 Hand-■man, Underrepresentation of Economic Groups on Federal Juries, at 207, 208.

Fourth, the standard for triggering supplementation was intentionally broad. Both the House and Senate Reports state that the affirmative obligation to supplement is triggered whenever there exists “any substantial percentage deviation” between the source lists and an ideal fair cross-section of the community, no matter how that deviation came to pass. S. Rep. at 17 (emphasis added); see also Hand-man, Underrepresentation of Economic Groups on Federal Juries, at 205 (“Apparently, the only inquiry relevant to supplementation is whether the prohibited deviation actually exists”). The definition of “substantial deviation” was left to the courts. S. Rep. at 17; 73 H.R. Rep. at 3.

Finally, neither the language of the Act nor its legislative history indicates that Congress intended for the finding of a Sixth Amendment violation to be a prerequisite to supplementation pursuant to § 1863(b)(2). After all, as described above, at the time the JSSA was passed, Sixth Amendment law was very narrowly drawn, focusing on intentional discrimination. It was not until Taylor in 1975 that Sixth Amendment law diverged from that approach and focused on concerns for a cross-sectional jury. 74

(3) The District of Massachusetts Resident List Exception

Two questions may be raised in connection with the application of the statutory language to this case. First, one could argue that Massachusetts has already met its burden of supplementation by choosing to use resident lists instead of voter lists. Second, the statute appears to be directed at the district court “Plan,” and not at the discretion of any given district court judge.

As described earlier, § 1863(b)(2) directs the district court to supplement “voter lists” where their use produces jury venires that insufficiently capture a cross-section of the community. Pursuant to that provision, in 1989, our district switched to resident lists. See supra note 18. In 1992, the JSSA was amended to reflect that change. See 28 U.S.C. § 1863(b)(2).

As to the first issue, the 1992 amendment (and Massachusetts’ choice of resident lists) should not be read to have fulfilled this Court’s obligation to ensure selection from a fair cross-section of the community. Section 1863(b)(2) refers to supplementing by using “some other source or sources of names,” implying that any single list, whatever it consists of, may not be sufficient. Moreover, as the legislative history discussed above suggests, the supplementation provision is concerned with achieving cross-sectional jury pools, no matter what the original source list or what the form of supplementation. Accordingly, the better interpretation of the statute is that the provision mandating supplementation of voter lists be read as mandating supplementation of resident lists in the case of the District of Massachusetts where those lists do not achieve the Act’s underlying purposes. Indeed, § 5(b) of the Jury Plan implies as much. It makes what is, in effect, a finding that “numbered local resident lists submitted annually [to the OJC] ... in accordance with Massachusetts General Laws Chapter 234A includes all registered voters, supplemented by all residents not registered to vote ...,” a finding that is plainly wrong in many Eastern Division cities and towns. Jury Plan § 5(b) (emphasis added).

This Court has previously had occasion to interpret the interplay between § 1863(b)(2)’s supplementation clause and Massachusetts’ use of resident lists: In U.S. v. Levasseur, 704 F.Supp. 1158 (D.Mass.1989), which predated the switch to resident lists, the defendants asked the court to require that potential jurors be drawn from resident lists rather than voter lists. In interpreting § 1863(b)(2), the court distinguished between supplanting and supplementing source lists, declaring that the District of Massachusetts would need express permission from the reviewing panel of the Jury Plan before it could supplant voter lists with resident lists. Id. at 1164. In other words, the Court found that switching to resident lists was not akin to supplementation pursuant to § 1863(b)(2), but a form of supplanting the statutorily provided source list with another.

Having established that Massachusetts’ use of resident lists does not in itself fulfill the duty to supplement under the JSSA, the question is whether an individual judge has the authority to order supplementation. Section 1863 also refers explicitly to a duty to supplement in the jury “Plan” rather than as a remedy ordered by an individual judge in an individual case. However, the remedy section of the JSSA, § 1867, addresses a finding by an individual court of a “substantial failure to comply with the provisions of this title,” calling for the court to “stay the proceedings pending the selection of a petit jury in conformity with this title.” 28 TJ.S.C. § 1867(d). To be sure, this language could suggest a stay in an individual ease until the reviewing panel formally amends the Plan, but such an interpretation is not the only one. Surely, certain kinds of limited remedies, namely those which do not fundamentally alter the “Plan” (the kind of “supplementation” and not “supplanting” that the Court referred to in Levasseur), are fully consistent with § 1867 and the Court’s supervisory authority. See infra Part III.B.2.e.

Indeed, the Massachusetts Plan on its face permits certain adjustments in individual cases. Section 11(d) of the Plan allows a court to direct the Clerk to draw a supplemental jury array from the master jury wheel to be added to the regular array as necessary when additional names are “needed.” Where the names from urban and minority areas are not accurate where sending out summonses to illusory addresses is nothing but an empty formulation, plainly additional names — of real people at real addresses — are needed in order to effect jury selection from a real cross-section of the community.

c. Have Defendants Proven a “Failure to Comply” Without Supplementation?

As described previously, the District’s overall rate of undeliverable summonses and nonresponses to summonses are troubling; for the years 2001 through 2003, 12.4% of summonses were returned by the post office to the federal Jury Administrator marked “undeliverable,” and another 12.2% of summonses received no response. In other words, for these years roughly one-quarter of all summonses mailed by the federal Jury Administrator using the names and addresses provided by the OJC were fruitless.

The District’s rates of nonresponses and undeliverables would be tolerable if they were uniform across the Eastern Division, but they are not. For example, among the group of summonses mailed to addresses in the Town of Lexington from 2001 through 2003, 3.7% were returned as undeliverable, and 5% were never returned (nonresponses). Meanwhile, among the group of summonses mailed to addresses in the City of Boston during that same period, 15.2% were returned as undeliverable, and 23% were never returned (nonres-ponses). There are blatant differences in the way cities and towns prepare the lists — how frequent, how careful — which further undermines minority representation.

The government highlights the difference between undeliverables and nonres- ponses. While the undeliverables involve wrong addresses, the nonresponses may involve correct addresses with noncoopera-tive residents, though the data is not clear. Regardless, while this distinction may have some resonance with respect to a constitutional analysis, it is largely irrelevant in the context of the JSSA’s affirmative obligations. As described above, the statute and its legislative history define the duty to supplement voter lists, even though citizens may choose not to vote. Likewise, there should be a duty to supplement resident lists, even when citizens choose not to respond to summonses.

On their face, these disparities undermine the JSSA’s policies of uniformity and randomness; representation of given communities is “haphazard” in the language of the legislative history. But they cut even deeper into the spirit of the Act, insofar as they contribute to the underrepresentation of African-Americans, not to mention urban and poor Eastern Division residents on Eastern Division jury venires.

For instance, for the Eastern Division’s 21 most heavily affluent and white towns, the average undeliverable rate was only 5.8% for the years 2001 through 2003. Meanwhile, for the three Boston neighborhoods that make up 76% of the city’s African-American population and 38.7% of the Eastern Division’s African-American population — Roxbury, Dorchester and Matta-pan — the undeliverable rate was 11.5% and the nonresponse rate was an alarming 35.6% for the years 2001 through 2003. These three urban neighborhoods are also heavily poor, with 23.7% of their combined population living at or below the poverty line. In contrast, the average poverty rate at the time of the 2000 Census for the Eastern Division as a whole was 9.3%.

Defendants also show that undercount-ing of residents is the greatest in Suffolk and Middlesex Counties, home to a majority of the Eastern Division’s African-American residents (roughly 64%). Indeed, in 2001, the resident list count of the voting-age population of Suffolk County differed from the Census estimate by negative 13.51%. 75

It is hard to imagine a situation more worthy of remedial action pursuant to § 1863(b)(2)’s supplementation provision. Admittedly, the federal court has no authority to order the city and town clerks to prepare their resident lists more painstakingly — the OJC-provided single numbered resident lists come to the federal court “as-ís.” But that does not absolve us; rather, the JSSA specifically mandates mechanisms for redressing deficient source lists.

d. Does Failure to Supplement the Resident Lists Amount to a Substantial Statutory Violation?

The government argues that, even if the federal Jury Administrator’s failure to supplement resident lists amounts to a violation of § 1863(b)(2)’s mandate, that violation is not “substantial.” I disagree. As stated above, “substantial” is measured “against the underlying principles of the Act.” Calabrese, 942 F.2d at 227; see also Brummitt, 665 F.2d at 528; Gurney, 393 F.Supp. at 701-02 (substantial noncompliance is “a patent violation of the spirit and letter of the Act”). By that measure, this violation is significant. It is not clerical, incidental, or inconsequential; rather, it strikes at the core of what the JSSA was intended to achieve. Cf. United States v. Nelson, 718 F.2d 315 (9th Cir.1983) (permitting a summoned juror who mistook his duty date to serve as a “volunteer” was technical violation); United States v. Tarnowski, 429 F.Supp. 783 (E.D.Mich.1977) (failure to abide by district’s jury plan requirement that the master jury wheel be emptied and refilled every two years was technical violation because time is not the essence of the Act); Gurney, 393 F.Supp. 688 (failure to follow district’s jury plan requirement that the names of residents from one county be transferred to a different division in order to comply with a division-modification order was technical violation).

Finally, failing to supplement deficient source lists is not trivialized by this Court’s finding that the underrepresentation identified by defendants’ data does not constitute a Sixth Amendment violation. If Congress had trusted that constitutional review would ensure truly cross-sectional jury pools, they would not have had reason to codify a fair cross-section ideal and enact mechanisms for its enforcement. Indeed, even if the longstanding pattern of underrepresentation of African-Americans in Eastern Division jury pools — a pattern that has been condemned by every court that has dealt with it, all the while finding it lawful — is not constitutionally actionable, the district court’s failure to take affirmative steps to improve the situation is actionable under the statute.

e. Supervisory Powers

While I believe that I have full authority to make certain of the adjustments that the defendants request as a matter of statute, and under this District’s Plan, there is another source of authority, namely this Court’s supervisory authority. “In the exercise of its supervisory authority, a federal court ‘may, within limits, formulate procedural rules not specifically required by the Constitution or the Congress.’ ” Bank of Nova Scotia v. United States, 487 U.S. 250, 254, 108 S.Ct. 2369, 101 L.Ed.2d 228 (1988), quoting United States v. Hasting, 461 U.S. 499, 505, 103 S.Ct. 1974, 76 L.Ed.2d 96 (1983). To be sure, my supervisory power is not unfettered; it “is applied with some caution even when the defendant asserts a violation of his own rights.” United States v. Payner, 447 U.S. 727, 734-35, 100 S.Ct. 2439, 65 L.Ed.2d 468 (1980).

The Supreme Court has explicitly dictated the three purposes underlying the use of the supervisory powers: “to implement a remedy for violation of recognized rights; to preserve judicial integrity by ensuring that a conviction rests on appropriate considerations validly before the jury; and finally, as a remedy designed to deter illegal conduct.” Hasting, 461 U.S. at 505, 103 S.Ct. 1974. Plainly, the remedies sought by these defendants are in response to a violation of “recognized rights.” Moreover, these remedies go to the very core of the system’s integrity. Nor do they conflict with any other constitutional or statutory provisions. Nova Scotia, 487 U.S. at 254, 108 S.Ct. 2369 (“even a sensible and efficient use of the supervisory power ... is invalid if it conflicts with constitutional or statutory provisions”) (internal quotations omitted).

Indeed, long before the JSSA and the development of the Sixth Amendment, the Supreme Court has regularly invoked the federal court’s supervisory authority to address jury selection issues, including discrimination claims. See Thiel v. Southern Pac. Co., 328 U.S. 217, 220-221, 66 S.Ct. 984, 90 L.Ed. 1181 (1946) (condemning the exclusion of “daily wage earners” from the venire, noting, “the choice of means” by which a jury is chosen fairly “rests largely in the sound discretion of the trial courts and their officers” so long as that judgment is “guided” by pertinent statutory provisions); Ballard v. United States, 329 U.S. 187, 192-93, 67 S.Ct. 261, 91 L.Ed. 181 (1946) (invoking federal court’s power of supervision over the administration of justice and reversing conviction of female defendant who was convicted by a jury consisting of no women).

In addition, for two defendants, death is a possible punishment, and for the three others, life imprisonment. The circumstances are all the more appropriate for exercising my supervisory powers to fashion a remedy for the Eastern Division’s unrepresentative jury pools. See. e.g., Fay v. People of New York, 332 U.S. 261, 287, 67 S.Ct. 1613, 91 L.Ed. 2043 (1947) (federal courts may exercise supervisory power over the selection of federal jurors “to reflect ... notions of good policy” and not simply constitutional mínimums); United States v. Whitley, 491 F.2d 1248, 1249 n. 1 (8th Cir.1974) (same).

f. Remedy

Defendants have moved for dismissal of the indictment, or, alternatively, for a stay while the District Court corrects the problem. See Corrected Second Declaration of Andrew A. Beveridge in Support of Defendant Branden Morris’ Motion To Dismiss, Defendants’ Exhibit 6, docket entry # 306, filed February 18, 2005; see generally Ellis and Diamond, Race, Diversity, and Jury Composition, at 1055-58.

I decline to dismiss because I believe that the statutory violation can be redressed with the measures I describe. Thus, I am effectively choosing to stay the jury selection proceedings, but the stay is a nominal one. The Federal Jury Administrator confirmed (in open court) that adequate remedial measures can be completed in time for the scheduled trials. 76 Defendants having been detained pending trial since 2002, it is imperative that a remedy is fashioned in a timely manner.

Accordingly, I invoke the statutory remedy provided in § 1867(d), as well as my supervisory authority over jury selection in my session. From this platform, I will instruct the federal Jury Administrator to conduct a targeted second mailing of summonses to account for the numbers of un-deliverables no matter where they occur. As defendants describe it:

The Jury Administrator plans to mail out approximately 2,000 juror summonses for each of the trials in this case. For all summonses returned to the court as ‘undeliverable,’ the same number of new summonses should be mailed to residents who live in the same zip code area as the undeliverable summons. The replacement summonses could be randomly drawn by computer programming. The Federal Jury Administrator should submit the names and address of the “undeliverables” to the OJC, and the OJC should instruct the cities and towns to remove these names and addresses from their resident lists. 77

Defendant Jonathan Hart’s Motion to Supplement The Juror Selection Process, docket entry # 382, filed July 15, 2005 at 1 [hereinafter “Defendants’ Motion to Supplement”].

With respect to nonresponses, I will not order a procedure, such as home visits, for determining whether these are real addresses. The administrative burden would be too substantial. Moreover, as I have suggested, such steps are unnecessary. Whether the reason for the nonresponse is that the address was inaccurate, or that the resident chose not to respond, is irrelevant. The issue is not the citizen’s choice to be included in the jury selection system; it is the defendant’s right to a fair cross-sectional jury pool. Thus, I will order “follow-up” questionnaires to be sent to the nonresponders, but if there continues to be no response, I will order the Jury Administrator to treat these addressees as undeliverables in the fashion described above.

I will also instruct the federal Jury Administrator to take other measures to help the OJC improve the accuracy of its single-numbered resident list; a dialogue that should have begun years ago when the District Court first became aware of the shortcomings of the resident lists.

I will also submit this opinion and Professor Abramson’s reports to the District Court, and the statutory reviewing panel to consider other options, including the “weighted mailing” and “under yield” approaches, for a more permanent solution.

While defendants’ proposal for weighted summonsing may well be a fitting approach for the District Court to consider in the future, it is not necessary in the ease at bar, in the light of the other remedies I have selected. In addition, since the “weighted mailing” or “under yield” approaches would involve adjustments to jury selection procedures, it may be more appropriate to submit this issue to other court decisionmakers. As Chief Judge Young announced when asked to invoke his supervisory powers to significantly modify the Jury Plan for selecting jurors, there should be reluctance for a single judge in a multi-judge district to unilaterally alter the standing jury plan in this fashion. See Levasseur, 704 F.Supp. at 1164 n. 10. In contrast, the remedy I have ordered, geographically-targeted second-round mailing of summonses to compensate for specific undeliverables and nonres-ponses, does not require altering the Plan at all, and thus does not present the prudential concerns faced by Chief Judge Young in Levasseur, 704 F.Supp. at 1164 n. 9. See Defendants’ Motion to Supplement.

g. Afterward: The Government’s Objections to the Proposed Remedy

After this decision had been submitted in draft form to the parties (and the public), the government noted its objections. Although the United States Attorney’s office agreed with the Court’s conclusions that there was no constitutional violation, that the Act’s proportionality requirement was not violated, that the weighted mailing approach should not be implemented by a single judge, it disagreed — sadly, rather vehemently — with even the limited remedy proposed here: It argued that whenever a summons is returned as “undelivered,” whenever there is no response after at least one subsequent mailing, this Court lacked the authority to send out additional summonses to others within the same zip code. Put otherwise, there is essentially nothing this Court can do (except the usual hand wringing) to ameliorate the unrepre-sentativeness of Eastern District juries in this capital case. The government’s position, troubling in its implications, is worth addressing in detail.

First, the government claims that only the Clerk, in conjunction with the Chief Judge, has the authority to implement even this remedy under the Jury Plan. Not so. The Chief Judge agreed that the steps I have taken are entirely consistent with the Jury Plan. (Transcript, August 31, 2005, at 71.)

Second, the government claims that the remedy I have proposed conflicts with Professor Abramson’s position. 78 Again, not so. Professor Abramson noted:

1. I agree with and support the Court’s proposed remedy. Under both 28 U.S.C. 1867(d) and the Court’s inherent supervisory powers over jury selection, the Court has the authority, and arguably the duty, to issue a remedy tightly and narrowly tailored to failures of jury selection in a particular case to conform to the policies of the Jury Selection and Service Act. This kind of narrowly tailored remedy, specific to a particular jury selection, is precisely what the Court has indicated it will order.

2. The proposed remedy is indeed geographically based, and does not target prospective jurors on the basis of their race. This is crystal clear because the Court has indicated it will send out additional summonses to replace every undeliverable summons, and every instance of nonresponse, no matter what zip code the undeliverable summons or nonresponse comes from. Indeed, the Federal Jury Administrator will have no knowledge of the race of persons to whom a second mailing is made.

3. The proposed remedy does not alter the District’s Jury Plan one iota and is thus different from a “weighted” jury selection. The Court is scrupulously following the Jury Plan and doing nothing more than attempting to fashion a remedy to achieve the paramount purpose of the Jury Plan, which is to summons persons randomly from a cross section of the community.

4. My initial report did not advise the Court that remedies such as these, or indeed weighted jury selection proposals, were a violation of the Jury Plan or the Jury Selection and Service Act. Rather, my report attempted to draw the Court’s attention to arguments that have been made on both sides of the issue as to whether weighted or stratified jury selection is consistent with the policy of random jury selection mandated by the Act and by the Plan.

Letter of Professor Abramson, dated September 1, 2005, document # 428.

Third, the government mischaracterizes the remedy ordered here, conflating it with the race conscious remedies that other courts have tried. This supplementation is not race-based; the second set of summonses will be mailed to addresses on the basis of their geographic location, and not on the prediction that they belong to prospective jurors of a particular race. 79 If a summons is returned from one of the more homogenous zip codes, a second summons will be sent to that area, without regard to the race of the addressee. No one will be overlooked. At the same time, “(a) key feature of this approach is that if a minority group clusters in particular locations”—as is largely true in the Eastern Division—“the system will correct for un-derrepresentation of that minority. The correction will occur automatically.” Ellis and Diamond, Race, Diversity, and Jury Composition, at 1057.

Fourth, a second round of summonsing does not jeopardize the randomness of the jury selection process, as randomness is defined by both the JSSA and the Jury Plan. See 28 U.S.C. § 1861; Jury Plan, at ¶ 5(c). The government maintains that these documents strictly define randomness as the method of selecting names from the Master Jury List to create the federal jury pool. Once the computer generates the federal list from the OJC’s Master Jury List, summonses must be sent out to the names on those lists and only those names, no matter how inaccurate they are. Any other procedure violates the law.

The government’s position is extraordinary. It elevates one concern of the Act over all others. The Act by its terms focuses not merely on the selection process'—that it be “at random”—but also on the source of the names, “a fair cross-section of the community.” 28 U.S.C. 1863(b)(3). The legislative history pointedly suggests that Congress was as concerned with the representativeness of the source lists, the “fair cross-section” issue, as with the means of selection. To make the argument that the Act’s purposes are fulfilled even if the source list is inaccurate, and summonses are sent to nonexistent persons, or the wrong addresses, is to adopt an empty formalism. It is a random selection from phantoms.

To the extent Congress was concerned with the means of selection, the focus was not statistical perfection. Indeed, the origins of the JSSA lie in attempts to dismantle the key man system. “Random” selection means the absence of any arbitrary attempt to include a particular class of persons. United States v. Davis, 518 F.2d 81 (10th Cir.1975) (“The essence of randomness ... is not number, but the absence of any arbitrary attempt to exclude a class of persons from the jury”).

Finally, as Professor Abramson notes, this remedy would in fact increase the randomness of the selection process because each person—each real person not each phantom—-would have more of an equal chance of receiving a summons. As Professor Abramson notes:

While there may be a certain strict logic to the government’s interpretation of the meaning of randomness, it is a position that would render an individual federal judge presiding over a jury selection virtually powerless to fashion a remedy, when as here in a particular case, there is evidence that the paramount policy of random selection which inspires both the Act and the Plan will not be met.

Abramson letter, supra.

The remedial procedure will be as follows:

The Jury Administrator will create a supplemental list of jurors from the names on the jury wheel that are not necessary for this year’s jury selection. (Each year there are extra juror names on the federal list that are not called up for the monthly venires.) He will send out additional summonses and questionnaires to individuals on that supplemental list to replace undeliverable or unresponded to summonses from the same zip code. Assuming these questionnaires are answered, the additional names will be merged with the names from the initial mailing, and then randomized into a single list.

Again, Professor Abramson summed it up best, “At some point, the life of the law is justice, not abstract logic, and I believe the Court’s carefully and limited remedy captures the better meaning of random jury selection in this case.” Abramson letter, supra. 80

IV. CONCLUSION

In light of the District Court’s failure to direct the federal Jury Administrator to supplement deficient resident lists, as is required by the JSSA, § 1863(b)(2), the jury selection of the defendants is stayed pursuant to § 1867(d) until the measures described above are implemented. The Court’s orders, however, are to be immediately implemented. The Court has every confidence that the remedial measures will be effected by the scheduled trial date. 81

One thing is dear: This Court cannot — ■ yet again — return to business as usual and cast a blind eye to real problems with the representation of African-Americans on our juries, and the crisis of legitimacy it creates.

SO ORDERED.

1 . This decision was first submitted to the parties (and the public) in draft form on August 23, 2005. This final version reflects the written and oral comments made by the parties after the draft decision was released. It also reflects comments made by the Court's expert, Professor Jeffrey Abramson, on September 1, 2005.

2 . An indictment was returned against Green, Morris, Jonathan Hart, Edward Washington, and Torrance Green (who will be referred to by his full name in this opinion) on July 17, 2002. A superceding indictment was returned on September 17, 2003. Counts One through Three of the superceding indictment charge all five with racketeering (18 U.S.C. § 1962(c)), racketeering conspiracy (18 U.S.C. § 1962(d)), conspiracy to murder, and murder in aid of racketeering (18 U.S.C. § 1959(a)(5)) during the period between June 2000 and September 2001. Counts Four through Seventeen charge specific individuals with various assaults and firearms offenses. In addition, the government filed a Notice of Intent to Seek the Death Penalty pursuant to 18 U.S.C. §§ 3591-92 against Green and Morris.

3 . The jury wheel is the electronic database that stores and randomly selects names of potential jurors. In the federal courts, the jury wheel is called the “master jury wheel.”

4 . At the outset, defendants additionally alleged violations of the Equal Protection Clause of the Fourteenth Amendment, but that ground for relief was dropped in subsequent pleadings.

5 . As an alternative remedy to dismissal, defendants move to supplement the jury wheel from which their petit jury will be drawn so that the racial composition of the wheel more closely reflects a fair cross-section of the community. Specifically, they move for: (1) a geographically weighted mailing of juror summonses to cure the underrepresentation of certain Eastern Division communities; (2) a stay, as provided in § 1867(a) of the JSSA, until a new jury wheel is assembled that complies with the Sixth Amendment and the Act; or (3) the implementation of other procedures for ensuring a representative jury, as laid out in Defendant Branden Morris’s Reply to the Government’s Additional Opposition to Defendants’ Motion to Dismiss [document #328], filed April 20, 2005, pp. 15-17. Defendants’ proposed alternative procedures are addressed infra in Part III.B.2.

6 ."Poor” is used in this opinion to refer to those persons living below the poverty line, as measured by the U.S. Census Bureau. The U.S. Census is the source for data regarding the numbers of poor persons living in certain geographic areas.

7 . It "not only hears factual allegations, as juries typically do, but it also weighs certain punishment factors, as judges typically do, to determine the defendant's sentence.” United States v. Green, Amended Memorandum and Order Re: Motions to Strike Notice of Intent to Seek the Death Penalty, dated June 6, 2005, pp. 6-7.

8 . As the Chair of The Jury Project of New York concluded in her introductory remarks to her commission: "Among minorities, a perception that they are not being called to serve in sufficient numbers exacerbates existing suspicions about whether the justice system works for minorities or is stacked against them.” Memorandum from Colleen McMahon, Chair, to Members of The Jury Project, Introductory Remarks 2 (August 13, 1993) (cited in Nancy J. King, Racial Jurymander-ing: Cancer or Cure? A Contemporary Review of Affirmative Action in Jury Selection, 68 N.Y.U. L.Rev. 707, 765 n. 210 (1993)).

9 . See letter of Professor Jeffrey Abramson dated September 1, 2005 (document # 428): "I agree with and support the Court's proposed remedy. Under both 28 U.S.C. 1867(d) and the Court's inherent supervisory powers over jury selection, the Court has the authority, and arguably the duty, to issue a remedy tightly and narrowly tailored to failures of jury selection in a particular case to conform to the policies of the Jury Selection and Service Act. This kind of narrowly tailored remedy, specific to a particular jury selection, is precisely what the Court has indicated it will order.”

10 . Transcript of Hearing dated August 31, 2005, at 71, "[Chief] Judge” Young has au thorized me to say ... "Judge Gertner and I have conferred and I ... am in agreement that these steps are entirely consistent with the Jury Plan.”

11 .The record includes: 1) three rounds of party briefing; 2) sixteen defense exhibits, thirty-four defense tables; 3) six declarations of defendants' jury expert, Professor Bever-idge; and 4) several boxes of completed questionnaires from city and town clerks. For the most part, the government concedes the accuracy of defendants' quantitative and qualitative data. To supplement the parties' paper briefing, I held hearings on defendants’ challenge on January 11, January 26, and March 24, 2005.

12 . After Professor Abramson submitted his report on April 22, 2005, I gave the parties more than two weeks to respond to his conclusions. Both parties responded, with the government commending the report's conclusions and defendants dissenting from them.

13 . Each district is required to formalize its jury selection procedures in written form pursuant to 28 U.S.C. § 1863(a). Accordingly, the Massachusetts District Court devised the Plan for Random Selection of Jurors ("the Jury Plan” or "the Plan”). The Plan was last revised in November 2000.

14 . This overview is laid out in: King, Racial Jurymandering, at 712.

15 . The District of Massachusetts, like judicial districts across the country, implemented a plan to govern jury selection procedures in the District. See supra note 9.

16 . The Eastern Division includes nine counties: Essex, Middlesex, Suffolk, Norfolk, Bristol, Plymouth, Barnstable, Dukes and Nantucket Counties.

17 . To a degree — and subject to court rules— law enforcement also chooses the venue (geographic area) for prosecuting the case.

18 . See Laura G. Dooley, The Dilution Effect: Federalization, Fair Cross-Sections, and the Concept of Community, 54 DePaul L.Rev. 79, 105-109 (2004)(hereinafter “Dooley, The Dilution Effect”)(“just as the minority vote gets diluted in at-large districting schemes,... the values of minority communities are more likely to be subsumed in juries drawn from larger federal districts than they would be in smaller, county-based state court juries''); see generally Darryl K. Brown, The Role of Race In Jury Impartiality and Venue Transfers, 53 Md. L.Rev. 107 (1994).

19 .In the death penalty context, the demographic effect of federalizing crime is especially important. The death penalty jury is asked to consider mitigating circumstances in determining whether the death penalty is justified, and to give content to that vague concept. Dooley, The Dilution Effect, at 105. Empirical research indicates that there are differences in the way African-American defendants are treated as the proportion of African-Americans on juries increases. David Baldus, et al., Racial Discrimination and the Death Penalty in the Post-Furman Era: An Empirical and Legal Overview, 83 Cornell L.Rev. 1638, 1724-25 (1998); see also William J. Bowers, et al., Death Sentencing in Black and White: An Empirical Analysis of the Role of Jurors’ Race and Jury Racial Composition, 3 U. Pa. J. Const. L. 171, 189 (2001) [hereinafter "Bowers, Death Sentencing ”] (finding the presence of African-American male jurors in cases involving African-American defendants and white victims substantially reduced the likelihood of a death sentence). One study found that the addition of one African-American male juror in a capital case made a stark statistical difference: "in the absence of black male jurors, death sentences were imposed in 71.9% of the cases, as compared to 42.9% when one black male was on the jury.” Dooley, The Dilution Effect, at 108, citing Bowers, Death Sentencing, at 193.

20 . Defendants have not argued that the government chose a federal forum precisely to affect the racial composition of the jury, an argument that may well raise Equal Protection concerns.

21 . This language is significant. It suggests an affirmative obligation for officials to ensure cross-sectional jury selection, rather than simply an obligation to stop purposeful discrimination. See Laura R. Handman, Case Comment, Underrepresentation of Economic Groups on Federal Juries, 57 B.U.L.Rev. 198, 205 (1977) [hereinafter "Handman, Underrep-resentation of Economic Groups on Federal Juries ”]. See infra Part III.B.2.

22 . Three years earlier, in 1989, the District amended the Jury Plan to replace voter lists with resident lists as the source of names for potential jurors.

23 . When the Act was amended in 1992, it was widely believed that using voter lists as the source of juror names caused paltry African-American representation in the jury venire, a belief that was supported by judges' anecdotal experiences of juries in their districts. See G. Thomas Munsterman & Paula Hannaford, Reshaping the Bedrock of Democracy: American Jury Refonn During The Last 30 Years, 36 No. 4 Judges' J. 5, 6 (1997). Indeed, when the Act was initially passed in 1968, its drafters did not contemplate whether voter registration lists were themselves representative of communities. As it turns out, they were not. See G. Thomas Munsterman & Janice T. Mun-sterman, The Search for Jury Representativeness, 11 Just. Sys. J. 59, 66 (1986); Cynthia A. Williams, Note, Jwy Source Representativeness and The Use of Voter Registration Lists, 65 N.Y.U. L.Rev. 590 (1990).

24 . Chapter 234A, § 10 requires that "each city and town shall make a sequentially numbered list of the names, addresses, and dates of birth of all persons who were seventeen years of age or older as of the first day of January of the current year and who resided as of the first day of January of the current year in such city or town.”

25 . "The cost of preparing the numbered resident list shall be paid by the city or town.” M.G.L. ch. 234A, § 10.

26 . Generally, the jury wheel is emptied and refilled periodically, usually every year or two. In the District of Massachusetts, the jury wheel is emptied and refilled annually.

27 . 28 U.S.C. § 1865(b) qualifies any person 18 years old or older who is a citizen of the United States and who has resided in the judicial district for one year, unless she cannot read, write and understand English sufficiently to fill out the juror qualification form; cannot speak English; has a mental or physical infirmity that would make her incapable of rendering satisfactory jury service; or has been convicted of a crime punishable by more than one year imprisonment and has not had her civil rights restored, or has such a charge pending.

28 .No data on the racial composition of the master jury wheel is available because race is generally not recorded on the numbered resident lists and is not part of the statutorily mandated information that cities and towns are required to collect under M.G.L. ch. 234A, § 10.

29 . Based on the return rate of summonses sent to prospective federal jurors from 2001 through 2003, defendants estimate that generally 12.4% of summonses were returned by the post office as ''undeliverable.” An additional 12.2% of summonsed persons never responded at all for reasons that can only be surmised.

For those who receive a summons — their summonses are not returned by the post office as ''undeliverable”- — but fail to complete and return the juror questionnaire, the federal Jury Administrator sends a follow-up letter. Those who fail to respond even after a second mailing are deemed "nonrespondents.” As described below, the rate of "undeliverables” and "nonrespondents” varies widely across the Division.

30 . Abramson emphasizes that, all told, we know the race of only 68% of the total number of persons summonsed for federal jury service from 2001 through 2003 (45,487 of the total pool of 67,157), after accounting for the number of persons who never responded, or responded without identifying their race. Obviously, there is no way of knowing the racial composition of the nonrespondents— that part of the pool from whom we have not heard at all. Abramson underscores this because he construes the Sixth Amendment pri-ma facie case very strictly — that it requires not simply a showing that official misfeasance contributes to some degree to the problem of minority underrepresentation, but a showing of precisely how much, to what degree. I am not convinced that the test is, or should be, that rigorous. In any event, for the purpose of computing the absolute disparity, the 3% figure is appropriate. The juries that will hear defendants' cases come from the pool of those who did respond, and we know that that pool is only 3% African-American.

Defendants use the term "available jury wheel” to describe the group of potential jurors who are summonsed, return a completed juror questionnaire, and identify their race. This designation is somewhat misleading because not every person returning the questionnaire (and thus becoming an "available” juror) identifies their race. The available jury wheel is somewhat larger than the pool of persons that defendants are actually using to calculate levels of underrepresentation. In 2001, 92.58% of persons returning questionnaires identified their race. For 2002 and 2003, the percentages are 91.83% and 83.49% respectively, averaging out to 88.9% of persons on the available jury wheel identifying their race from 2001 through 2003. Def. Ex. 3, Amended Table 4. So long as one assumes that African-Americans are not disproportionately over-represented in this relatively small portion of the questionnaires with no race identified, then it is reasonable for defendants to extrapolate from the racial makeup of the 88.9% of the persons identifying their race on returned questionnaires to 100% of names on the available jury wheel. See, e.g., Shari Seid-man Diamond, "Reference Guide on Survey Research,” in Reference Manual on Scientific Evidence at 245 ("[Rjesponse rates of 90% or more are reliable and generally can be treated as random samples of the overall popula tion.”); see also Abramson, Report, p. 18 n. 13.

31 . The parties agree that the United States Census provides the best data as to the percentage of African-Americans of voting age in the Eastern Division. See, e.g., United States v. Rioux, 930 F.Supp. 1558, 1565 (D.Conn.1995) ("census data provides the most useful information” as to voting-age population). While it might be preferable to start with more refined information about the racial composition of the jury-eligible population, such data is rarely available. Id. However, even the Census data has been criticized as undercounting African-Americans, as well as other minorities. See Nathaniel Persily, Color by Numbers: Race, Redistricting, and the 2000 Census, 85 Minn. L.Rev. 899, 903 (February 2001).

32 . Death-qualification is the process by which potential jurors in a capital case are questioned during voir dire about their attitudes toward the death penalty. The government has the right to strike potential jurors who are unequivocally opposed to the death penalty. See Witherspoon v. Illinois, 391 U.S. 510, 520, 88 S.Ct. 1770, 20 L.Ed.2d 776 (1968).

33 .In United States v. Gilbert, 98-cr-30044-MAP (D. Mass, filed Nov. 19, 1998), a capital case, among the 203 jurors who were questioned, only eight were African-American. Six of the eight opposed the death penalty (75%) while two favored the death penalty in only special circumstances (25%). Ultimately, no African-American jurors were seated. The result was the same in the district's other recent capital case, United States v. Sampson 01-cr-10384-MLW (D. Mass, filed Oct. 25, 2001), wherein only 23 of the 498 jurors who completed questionnaires were African-American (4.6%). Of those potential African-American jurors, ten opposed the death penalty (43.5%), one was in favor of the death penalty (4.3%), and ten were neutral (43.5%). Again, no African-Americans were seated on the capital defendant’s petit jury. See Memorandum and Order Re: Severance/Bifurcation of Guilt and Punishment Juries, docket entry # 193, dated July 7, 2004, pp. 37-38.

34 . Notably, the Massachusetts Governor's Council on Capital Punishment recently drafted a report recommending that capital defendants be given the choice to bifurcate their juries. See Massachusetts Governor's Council on Capital Punishment, Final Report, available at: http:/ /www.mass.gov/Agov2/docs/5-3-04MassDPReportFinal.pdf. In so recommending, the Council was compelled in part by data indicating that death-qualification significantly dilutes the diversity of jury pools.

35 . For instance, the questionnaire asked: when was the last time Census forms were mailed to the residents of the city or town; how does the city or town determine what names and addresses to mail the Census forms to; and what does the city or town do about Census forms that are not returned?

36 .With approval from the Court, the federal Jury Administrator supplied both parties with electronic copies of the master jury wheels for the years 2001, 2002 and 2003; all of the jury pools from 2001 through 2003 (57 pools); and all of the grand jury pools from 2001 through 2003 (7 pools).

37 . As described below in section III. B.(2), official misfeasance need not be shown in connection with a statutory "substantial failure to comply” case.

38 . New Bedford responded to the questionnaire by stating that it "has not done a census since 1999.”

39 . For instance, if a group formed 10% of the population but only 6% of the jury wheel, there would be an absolute disparity of 4%.

40 . Indeed, modem Sixth Amendment analysis is rooted in cases in which women were explicitly treated differently in jury selection for relatively benign reasons, namely administrative convenience. See Taylor, 419 U.S. at 522, 95 S.Ct. 692 (rejecting Louisiana requirement that a woman could not serve on a jury unless she filed a written declaration of willingness to do so); Duren, 439 U.S. at 357, 99 S.Ct. 664 (rejecting Missouri's automatic exemption for any woman requesting not to serve as a juror).

I say "modern” Sixth Amendment analysis because the fair cross-section theory as we understand it cannot be explicitly found in the Constitution. At the time of its drafting, juries consisted of white, male, and propertied citizens.

41 . This measurement is calculated by dividing the absolute disparity percentage by the percentage of the group in the population.

42 . Absolute disparity analysis yielded a disparity of 2.97%, and comparative disparity analysis yielded a disparity of 60.9%. Id. at 10, 11 & n. 10.

43 . Scholarly and judicial criticism of the absolute disparity approach has bourgeoned over the years. See, e.g., Detre, Measuring Underrepresentation, at 1921.

44 . Professor Abramson agrees. He notes,

[A]bsolute disparity figures are not self-interpreting and the significance of the numbers, in terms of real life effects on the right of defendants [to a representative jury] ... can only be grasped by asking how any particular drop-off in percentage representation of African-Americans affects the con-stitulional and statutory imperative that petit juries should be as representative as random selection from a fair cross-section of the voting-age population permits.

Abramson, Report, at 19.

45 .The Royal court acknowledged that the First Circuit used comparative disparity analysis in LaRoche v. Perrin, 718 F.2d 500 (1st Cir.1983), which concerned the alleged un-derrpresentation of persons 18 to 34 years old in the jury pool. It distinguished LaRoche, however, on the following ground: "[La-Roche ] did not adopt the comparative disparity analysis to deal with ... the situation in which the group allegedly underrepresented forms a very small proportion of the total population.” Royal, 174 F.3d at 7 (quoting Hafen, 726 F.2d at 24 n. 3). The Royal court also noted that "[a] small variation in the figures used to calculate comparative disparity can produce a significant difference in the result. .." Id. at 7-8 (quoting Hafen, 726 F.2d at 24).

46 . Other courts have been less rigid in applying absolute disparity analysis. In United States v. Weaver, for instance, the Third Circuit noted that "figures from both methods [absolute and comparative disparity] inform the degree of underrepresentation, [and therefore] we will examine and consider the results of both in order to obtain the most accurate picture possible.” 267 F.3d 231, 243 (3rd Cir.2001). Particularly apt is the rationale employed by a district court in Alaska, which followed the Ninth Circuit’s directive to use absolute disparity but declined to apply it in a vacuum:

In this court’s view, the absolute disparity test cannot reasonably be applied without some regard for the representation of the particular distinctive group in the total population. For example, an absolute disparity of 7.7% would be far more troubling when dealing with a distinctive group than it would be if the group made up 15% of the total population. Thus, while mindful of the Ninth Circuit’s teaching on the need to avoid 'exaggeration’ introduced when comparative disparity is used rather than absolute disparity, in this court's view comparative disparity can be used as an adjunct to absolute disparity as a means for assuring that the absolute disparity test is not applied in a vacuum.

United States v. Pleier, 849 F.Supp. 1321, 1329 (D.Alaska 1994).

47 . For a detailed analysis of both statistical measures, see generally Detre, Measuring Un-derrepresentation.

48 . Binomial distribution is a statistical distribution model that describes such phenomena as throwing dice. For a thorough description, see Detre, Measuring Underrepresentation, at 1918 n. 26.

49 . Defendants also argue that the critical circumstances of this case distinguish it from First Circuit precedent, which did not deal with capital cases. As Justice Stevens has noted, Supreme Court "opinions have repeatedly emphasized that death is a fundamentally different kind of penalty from any other that society may impose.” Harris v. Alabama, 513 U.S. 504, 516, 115 S.Ct. 1031, 130 L.Ed.2d 1004 (1995) (Stevens, J., dissenting). That difference means that capital proceedings "require[] a correspondingly greater degree of scrutiny.” California v. Ramos, 463 U.S. 992, 998-99, 103 S.Ct. 3446, 77 L.Ed.2d 1171 (1983). Such scrutiny is especially appropriate in this case, where African-American defendants face the death penalty in federal court for what are essentially state crimes.

50 . SDT, for instance, has been criticized as an inappropriate measure of jury disparities because it is based on random selection. United States v. Rioux, 97 F.3d 648, 655 (2d Cir.1996) ("It is illogical to apply a theory based on random selection when assessing the constitutionality of a qualified wheel. By definition, the qualified wheel is not the product of random selection; it entails reasoned disqualifications based on numerous factors”). Disparity of risk, meanwhile, is disfavored because it has yet to garner approval from any court. See Commonwealth v. Arriaga, 438 Mass. 556, 566, 781 N.E.2d 1253 (2003).

51 . The First Circuit is surely not alone in its approach to this issue. See United States v. Biaggi, 680 F.Supp. 641, 655 (S.D.N.Y.1988) (4.7% absolute disparity found permissible), aff'd, 909 F.2d 662, 678 (2d Cir.1990), ce rt. denied, 499 U.S. 904, 111 S.Ct. 1102, 113 L.Ed.2d 213 (1991); United States v. Musto, 540 F.Supp. 346, 356 (D.N.J.1982) (absolute disparity of 5.4% found permissible), aff'd sub nom., United States v. Aimone, 715 F.2d 822 (3d Cir.1983), ce rt. denied, 468 U.S. 1217, 104 S.Ct. 3585, 82 L.Ed.2d 883 (1984); Bryant v. Wainwright, 686 F.2d 1373, 1377-78 (11th Cir.1982) (absolute disparity of 7.4% found permissible), cert. denied, 461 U.S. 932, 103 S.Ct. 2096, 77 L.Ed.2d 305 (1983); United States v. Hawkins, 661 F.2d 436, 442 (5th Cir.1981) (absolute disparity of 5.45% found permissible), cert. denied, 456 U.S. 991, 102 S.Ct. 2274, 73 L.Ed.2d 1287 (1982); United States v. Clifford, 640 F.2d 150 (8th Cir.1981) (absolute disparity of 7.2% found permissible); United States ex rel. Barksdale v. Blackburn, 639 F.2d 1115, 1126-27 (5th Cir.1981) (absolute disparity of 11.5% found permissible), ce rt. denied, 454 U.S. 1056, 102 S.Ct. 603, 70 L.Ed.2d 593 (1981); United States v. Armstrong, 621 F.2d 951 (9th Cir.1980) (abso lute disparity of 2.8% found permissible); United States v. Potter, 552 F.2d 901, 906 (9th Cir.1977) (absolute disparity of 2.7% found permissible); Thompson v. Sheppard, 490 F.2d 830, 832-33 (5th Cir.1974) (absolute disparity of 11.0% found permissible), cert. denied, 420 U.S. 984, 95 S.Ct. 1415, 43 L.Ed.2d 666 (1975).

52 . Clearly, the 10% rule adopted by some courts is a contrivance, and one based on faulty precedent. A number of courts have cited Swain v. Alabama, 380 U.S. 202, 85 S.Ct. 824, 13 L.Ed.2d 759 (1965), to buttress the use of the 10% figure. See e.g., United States v. Maskeny, 609 F.2d 183 (5th Cir.1980). In Swain, the Court held that "[it] cannot say that purposeful discrimination based on race alone is satisfactorily proved by showing that an identifiable group in a community is underrepresented by as much as 10%.” Swain, 380 U.S. at 208-09, 85 S.Ct. 824. But "purposeful discrimination” is not part of the fair cross-section analysis. The Fifth Circuit conceded as much but nevertheless adopted the 10% rule:

The Supreme Court in Swain v. Alabama held that underrepresentation by as much as ten percent did not show purposeful discrimination based on race. We recognize, however, that Swain was an equal protection case where purposeful discrimination must be shown and that the Court in Duren stated that a defendant need not show discriminatory purpose for the Sixth Amendment violation. The Court in Duren, however, discussed the statistical discrepancy needed to make out an equal protection violation along with its discussion of the disproportion that demonstrates a Sixth Amendment violation. Thus, while the Court stated that statistical evidence is used to prove different elements in Equal Protection and Sixth Amendment claims, it did not indicate that the necessary amount of disparity itself would differ.

Maskeny, 609 F.2d at 190.

53 . As one scholar described it, the question is whether the underrepresentation was "inherent in the system used, rather than a product of random factors on one particular jury veni-re." Cynthia A. Williams, Jury Source Representativeness and the Use of Voter Registration Lists, 65 NYU L.Rev. 590, 617 (1990) (emphasis added); see also, Duren, 439 U.S. at 364-66, 99 S.Ct. 664. An "individual instance of underrepresentation might be a coincidence, whereas a pattern will betray a systematic procedural abuse." James H. Druff, The Cross-Section Requirement and Jury Impartiality, 73 Cal. L.Rev. 1555, 1565 (1985) (emphasis added).

54 . These calculations are based on the available jury wheel because the racial composition of the master jury wheel is unknown. Notably, the government posits that data regarding the racial composition of the available jury wheel is irrelevant to a fair cross-section challenge. The government is correct that defendants cannot use the racial composition of the available jury wheel to extrapolate back to the racial composition of the master jury wheel. In Royal, for instance, both the district court and the First Circuit concluded that calculations about racial disparities in the available jury wheel likely overstated the absolute disparity on the master jury wheel. 174 F.3d at 10 n. 11, 7 F.Supp.2d 96, 102 (D.Mass. 1998). But the government is certainly incorrect to argue that this Court cannot rely on data about the racial composition of the available jury wheel. In Royal itself, despite some misgivings about the accuracy of the data, the district court found it proper to examine data about under-representation in the available jury wheel. 7 F.Supp.2d at 103-04 (''[T]he number properly used in the disparity calculations is not the number of black persons in the Master Jury Wheel (a number that can only be estimated), but the number of black persons actually responding to summonses.”). See also Pion, 25 F.3d at 22-23 (comparing census data on proportion of Hispanic voting-age population to percentage of those who identified themselves as Hispanic on jury questionnaires); Hafen, 726 F.2d at 23 (relying on percentage of summoned prospective jurors who answered race question on questionnaire); United States v. Levasseur, 704 F.Supp. 1158, 1161 (D.Mass.1989) (calculating racial gap on the "qualified jury wheel pool”).

55 .For example, Needham's efforts are aggressive: it begins with a street listing of addresses in the town, "which is then updated by the Engineering and Building Departments each time there is a new street or house built and the occupancy permit has been issued.” After the initial mailing, Needham does a second mailing. Some towns with similar demographics as Needham conduct third mailings, telephone calls, and even door to door visits. The Town of Sudbury reports that "we hand deliver to nursing homes and assisted living facilities.” Meanwhile, Lynn and Brockton — more racially and economically heterogenous towns — do no more than mail Census forms annually; they conduct no follow-up at all.

56 . The U.S. Census Bureau is the source for demographic data at the county, city, town and zip code levels.

57 . For this analysis, defendants matched the addresses on the master jury wheels for the years 2001, 2002 and 2003 with their corresponding zip codes.

58 . In voting rights cases, for example, courts frequently struggle with this problem; they have data only about the overall vote in a district and they seek to know whether whites and minorities voted for the same or for different candidates. See, e.g., Thornburg v. Gingles, 478 U.S. 30, 52, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986); Georgia v. Ashcroft, 195 F.Supp.2d 25, 69 (D.D.C.2002).

59 . A comparison of Boston and Dover illustrates the two ends of the spectrum: African-Americans constitute 0.4% of Dover’s population. Dover’s median household income is $141,818. Between 1999 and March 2000 (15 months), 9.3% of Dover’s population changed residences. In contrast, African-Americans constitute 25.3% of Boston's population. Boston’s median household income is $39,629. Between 1999 and March 2000, a whopping 24% of Boston's population changed residences.

60 . The literature does suggest that minorities disproportionately fail to respond to jury summonses — perhaps because they choose not to do so. See Hiroshi Fukarai et al., Race and the Jury: Racial Disenfranchisement and the Search for Justice 54 (1993) (minorities disproportionately do not respond to jury summonses since they hesitate to be part of a system they consider unfair, or perhaps because jury service does not rank high in their lives). Defendants disparage this conclusion, arguing that there is no evidence on the record or any other empirical reason to assume that African-Americans are more likely to ignore jury summonses.

61 . More specifically, the OJC leaves it up to each city and town to decide whether some or all of the listed inactive voters may still reside locally, and thus should remain on the voter list.

62 . In accordance with this provision of the Act, this district's Jury Plan at paragraph 7 specifies that "random selections of names from the source list for inclusion in the master wheel by data computer personnel must ensure that each county within the jury division is substantially proportionally represented in the master jury wheel.” In response to questions put by defendants, the OJC outlined the actual procedures used to generate a veni-re list for the Eastern Division that meets the substantial proportionality requirement: The federal Jury Administrator informs the OJC about the percentage of names on the numbered resident lists required to create the master jury wheel for the division (approximately 1%). The OJC then uses internal computer processes to generate random numbers and randomly select the required percentage of residents from each individual city and town in the relevant federal division. Ultimately, the names are shuffled and merged in a file for the federal court — that file is not organized according to city and town.

63 . The master jury wheel is constructed by taking one percent of the names from the resident lists submitted by the cities and towns. As long as the computations are correct, county percentages on the wheel are presumably the same as county percentages on the resident lists.

64 . Defendants also argue that the failure of several cities and towns to update their resident lists annually or accurately amounts to a “substantial failure to comply” with the Act. In so arguing, defendants rely on the following language in the Act at § 1863(b)(2): "The plan for the District of Massachusetts may require the names of prospective jurors to be selected from the resident list provided for in chapter 234A, Massachusetts General Laws, or comparable authority." Defendants point to evidence in the record that some cities and towns have failed to conduct surveys of their residents annually as required under Mass. Gen. L. ch. 234A, § 10. Several city and town clerks conceded that they fail to compile resident lists that count "all persons" seventeen years and older, as required by chapter 234A, because they lack the budgetary and personnel resources.

I cannot address the failure of state and city officials to comply with the Massachusetts statutes. I can only address the obligations of federal officials to do what is necessary to supplement the flawed lists to ensure a jury pool drawn from a fair cross-section of the population.

65 . This aspect of the Plan refers specifically to the use of data processing systems in generating names of individuals to be sent summonses, which could include a program for "weighted mailings” as the defendants request. See infra Part III.B.2. It does not on its face speak to multiple mailings to compensate for nonresponses and undeliverables in a given case, but given the statute’s broad mandate, is not inconsistent with that approach.

66 . 28 U.S.C. § 1863(a) requires that the Plan for Random Jury Selection be approved by a "reviewing panel consisting of the members of the judicial council of the circuit and either the chief judge of the district.. .or such other district judge ... as the chief judge may designate.”

67 . The defendants propose that federal officials check the names and addresses with the U.S. Postal Service and the Registry of Motor Vehicles and thereafter attempt to telephone each nonrespondent. If the officials are unable to determine whether the nonrespondent actually lives at that address, an official should visit the putative nonrespondent. If the nonrespondent does not live at the address, the summons should be classified as undeliverable. If the potential juror does live at that address, multiple questionnaires and summonses should be sent to him or her.

68 . Abramson described defendants' proposed "weighted” or "stratified” random sampling as follows:

Dr. Beveridge has compiled data showing how many names and addresses from a particular zip code appear on the actual Master Jury Wheel. He compares this to an Ideal Master Wheel where each zip code had its rightful proportion of names on the Master Wheel according to its share of the Census-measured voting-age population of the Eastern Division. He then calculates the over or under representation for each zip code on the Actual Wheel in comparison to the Ideal Wheel. On the basis of this information, he calculates for each zip code how to weight the mailing of summonses so as to proportionally reduce the chances that someone in an overrepresented zip code will receive one. By adjusting the summonses in this way, the hope is that the end result will be the summonsing of a pool of jurors that fairly reflects the voting-age population of the Eastern Division.

Abramson, Report, at pp. 59-60.

Defendants initially proposed implementing a less drastic method of weighted summonsing. In their memorandum filed on January 5, 2005, defendants proposed the following method, as described by Ellis and Diamond:

[Sjuppose that a sample of names is drawn from the list of potential jurors and a qualification questionnaire is sent to each name on the list. If a particular political district or ZIP code X accounted for 4% of the mailed questionnaires, but only 2% of the qualified jurors who appeared in the courthouse were from district or ZIP code X, that 'under-yield' would be adjusted in the next mailing. The mailed questionnaires to district or ZIP code X would represent 8% of the total mailed, with a predicted yield of 4% that would reflect the actual distribution in the initial list.

Defendant Branden Morris' Motion To Dismiss The Superseding Indictment On Account Of The Racial Composition Of The Master Jury Wheel, Or Alternatively, To Supplement The Master Jury Wheel, Docket entry # 255, filed January 5, 2005 (quoting Leslie Ellis and Shari Seidman Diamond, Race, Diversity, and Jury Composition: Battering and Bolstering Legitimacy, 78 Chi-Kent L.Rev. 1034, 1056 (2003)) [hereinafter "Ellis and Diamond, Race, Diversity, and Jury Composition ”].

69 . Drafters of the JSSA expressly eliminated the need "to prove prejudice as a condition of judicial intervention when substantial noncompliance with the act is established.” H.R.Rep. No. 1076, 90th Cong., 1st Sess. 13. The prejudice standard was seen as too burdensome. Id.

70 . "[T]he alleged violations must be weighed against the underlying principles of the Act.” United States v. Calabrese, 942 F.2d 218, 227 (3rd Cir.1991) (internal citations omitted); see also United States v. Brummitt, 665 F.2d 521, 528 (5th Cir.1981) ("[djetermining substantial compliance requires weighing the alleged violation against the goals of the Act”); United States v. Gurney, 393 F.Supp. 688, 701-702 (M.D.Fl.1974) (substantial noncompliance is “a patent violation of the spirit and letter of the Act”).

71 . See generally Handman, Underrepresentation of Economic Groups on Federal Juries, at 200-02.

72 . The hearings reflect a debate between those who believed that jurors should have a certain level of competence, and civic interest, and those who believed any substantial competency screening would undermine the cross-sectional goal, as the "key-man” system had done. S. Rep. at 18, 22.

73 . The Senate Report explained: "The voting list need not perfectly mirror the percentage structure of the community. But any substantial percentage deviations must be corrected by the use of supplemental sources. Your committee would leave the definition of 'substantial' to the process of judicial decision.” S. Rep. at 17.

74 .Two courts, notably the Fifth and Second Circuits have recognized a duty to supplement their source lists where a cognizable class is being substantially underrepresented in its jury pools, regardless of whether the source list was implemented in a discriminatory fashion. United States v. Jenkins, 496 F.2d 57, 65 (2d Cir.1974) (The JSSA goes beyond the “prohibition of 'intentional distortions’ and provides that in certain cases affirmative measures must be undertaken to ensure that ‘juries are selected at random from a fair cross-section of the community;’ ” minor deviations from a fully accurate cross-section are permitted); United States v. Goff, 509 F.2d 825 (5th Cir.1975) (adopting the standard spelled out in the lower court opinion, United States v. McDaniels, 370 F.Supp. 298 (E.D.La.1973)). Neither, however, concluded that the supplementation requirement was required by the statistics presented.

Moreover, Jenkins and Goff do not represent the majority approach. Many courts continue to wrongly elide the Equal Protection Standard and the statutory standard, limiting supplementation to the underrepresentation caused by the intentional and systematic exclusion of a cognizable class. See United States v. Diggs, 522 F.2d 1310, 1316 (D.C.Cir.1975); United States v. Freeman, 514 F.2d 171, 173 (8th Cir.1975); United States v. Lewis, 472 F.2d 252, 256 (3d Cir.1973) (dictum); United States v. Ross, 468 F.2d 1213, 1216 (9th Cir.1972); United States v. James, 453 F.2d 27, 29 (9th Cir.1971). That approach is inconsistent with current Sixth Amendment law, and the language and legislative history of the JSSA. See Handman, Underrepresentation of Economic Groups on Federal Juries, at 203.

75 . While urban residents may not be a cognizable class for the purpose of the Sixth Amendment, their underrepresentation deprives the jury pool of important perspectives, particularly where the jury will be analyzing urban street crime. It is telling that, in the decade I have served as a judge, I have heard many potential jurors from the suburban communities surrounding Boston state that they are "afraid” to come into the city, even for jury service.

76 . Transcript, August 31, 2005, at pp. 63-64.

77 . As defendants note, the OJC does not presently submit ''undeliverable” names and addresses to the cities and towns to help them improve their resident lists. My direction that the OJC "instruct” the cities and towns to update and improve their lists is merely advisory, since, as discussed supra, I have no authority to "instruct” the OJC or the city and town clerks to do anything with respect to their chapter 234A resident lists. Nevertheless, encouraging state authorities to improve their resident lists through an ongoing dialogue with the federal Jury Administrator is worthy of a strongly worded recommendation.

78 . The government, of course, objected to the appointment of Professor Abramson as the jury expert. They have obviously reversed their position in the most recent pleading in which they gave great prominence to what they reported to be his views. I can only assume that in the light of Professor Abram-son’s recent letter, wholly endorsing this Court’s remedy, the government will do another about face about the value of his expertise.

79 . In this important way, this remedy is distinct from the race-based remedy prohibited by the Sixth Circuit in United States v. Ovalle, 136 F.3d 1092 (6th Cir.1998). In Ovalle, the court reviewed a district court plan that adjusted the racial composition of the jury veni-re by eliminating potential jurors on the basis of their race (white) and replacing them with others on the basis of their race (African-American). The court found such a one-to-one replacement a violation of the JSSA's prohibition on exclusion from jury service “on account of race, color, religion, sex, national origin, or economic status.” 28 U.S.C. § 1862.

80 . On other issues raised by the government in a brief that ranged from the disingenuous to the disrespectful: This Court did not make a "fundamental error” in ''ignoring] the Act’s mandate that randomness must be measured within the Eastern Division as a whole, and not within political or other subdivisions.” The statute refers to a fair cross-section "of the persons residing in the community in the district or division wherein the court convenes.” 28 U.S.C. § 1863(b)(3) (emphasis added). By substituting a summons to a bad address with a another summons from the same area, a summons with a presumably accurate address, the Court maximizes the chances that "persons” will have an equal opportunity for access to this jury.

Second, this Court did not "fail[ ] to recognize that the Act and the Plan measure randomness at the time of initial selection of names from the source list and the Master Wheel.” The Plan suggests that the selection of names must come from a "complete source list database.” The government ignores the word "complete." The resident lists are far from complete. Indeed, in describing the selection process, the Plan notes that it “may be accomplished by a purely randomized process,” as if there were other alternatives. Jury Plan ¶ 7(a) (emphasis added). In the final sentence of § 7(a) the Plan describes the selection of names as deriving from multiple areas, "the source list,” i.e. the resident lists, and the "master wheel.” And in the same section it requires that whatever procedure is used — source lists and the selection process— "must insure that the mathematical odds of any single name being picked are substantially equal.” The concern for “source lists” in the Act, the legislative history and the Plan, the concern that "persons” have an equal opportunity to serve, belies the government’s argument.

Third, this Court is not "rewriting" the Act and "selectfing] jurors in any manner it wishes” merely upon a determination that it is "necessary to foster the policy” of the Act. The Court’s remedy here is entirely consistent with the language of the Act, its purposes, and the Plan and it could not be more narrowly focused as the Court's expert and the Chief Judge have acknowledged.

Fourth, while the "government is ... at a loss to understand how this Court could find that the use of Massachusetts resident lists (which it calls deficient) is a substantial violation of the Act” in the light of the First Circuit’s decision in United States v. Royal, 174 F.3d 1 (1st Cir.1999), the explanation is clear. It is all about facts, a record, witnesses, and data. The record in the instant case — about the patent deficiencies in the resident lists— distinguishes this case from those that have preceded it. While the government may have no problem ignoring the extraordinary return rates in our urban areas, this Court cannot. Indeed, it is the Court that is at a loss to understand how the government cannot see past its own arguments to fully appreciate its obligations to the citizens of this state.

81 . See Defendants’ Motion to Supplement.