Quinn v. City of Boston

325 F.3d 18 Joseph E. QUINN, et al., Plaintiffs, Appellants, v. CITY OF BOSTON, et al., Defendants, Appellees, National Association for the Advancement of Colored People, Intervenor, Appellee. No. 02-1727. United States Court of Appeals, First Circuit. Heard November 5, 2002. Decided March 27, 2003. COPYRIGHT MATERIAL OMITTED COPYRIGHT MATERIAL OMITTED COPYRIGHT MATERIAL OMITTED COPYRIGHT MATERIAL OMITTED Harold L. Lichten, with whom Pyle, Rome, Lichten & Ehrenberg, P.C. was on brief, for appellants. Christine M. Roach, with whom Roach & Carpenter, P.C. and Merita A. Hopkins, Corporation Counsel, were on brief, for defendants-appellees. Toni G. Wolfman, with whom Foley Hoag LLP, Nadine M. Cohen, Maricia Woodham, and Lawyers Committee for Civil Rights Under Law were on brief, for intervenor-appellee. Before SELYA, Circuit Judge, STAHL, Senior Circuit Judge, and LIPEZ, Circuit Judge. SELYA, Circuit Judge. — 1 — For over a quarter of a century, the hiring of firefighters in the City of Boston (the City) has taken place in the albedo of a federal court consent decree designed to remedy the effects of past discrimination against African-Americans and Hispanics. On April 11, 2001, five candidates for employment (the Candidates) brought suit in the federal district court alleging that the City had discriminated against them on the basis of race when hiring new firefighters in the fall of 2000. The City defended its hiring practices as compliant with, and compelled by, the terms of the consent decree. The district court granted summary judgment in the defendants' favor. See Quinn v. City of Boston, 204 F.Supp.2d 156 (D.Mass.2002). The Candidates appeal. For the reasons that follow, we reverse the judgment and remand to the district court for further proceedings consistent with this opinion. I. BACKGROUND — 2 — In considering an appeal from a grant of summary judgment, this court, like the trial court, normally will "view the entire record in the light most hospitable to the party opposing summary judgment, indulging all reasonable inferences in that party's favor." Griggs-Ryan v. Smith, 904 F.2d 112, 115 (1st Cir.1990). Here, however, the facts upon which our decision turns are undisputed. — 3 — In the early 1970s, two suits were brought against a number of municipalities subject to the Massachusetts Civil Service law (now codified at Mass. Gen. Laws ch. 31, §§ 1-77). The suits alleged that the municipalities engaged in discriminatory recruitment and hiring practices whilst staffing their respective fire departments. These actions culminated in the entry of an omnibus consent decree that influenced the manner in which the affected municipalities could recruit and hire firefighters. Boston Chapter, NAACP, Inc. v. Beecher, 371 F.Supp. 507, 520-23 (D.Mass.1974) (Beecher I). The decree was affirmed on appeal. Boston Chapter, NAACP, Inc. v. Beecher, 504 F.2d 1017, 1028 (1st Cir.1974) (Beecher II), cert. denied, 421 U.S. 910, 95 S.Ct. 1561, 43 L.Ed.2d 775 (1975). It has been in effect for nearly thirty years. — 4 — The Beecher decree — courts consistently have used that phrase to include the original decree and subsequent orders entered to fine-tune it, see, e.g., Mackin v. City of Boston, 969 F.2d 1273, 1274 (1st Cir.1992), and we emulate that example — circumscribed the hiring of firefighters in much of Massachusetts over the ensuing years. Its history up to 1992 is well chronicled in the case law. See, e.g., id. at 1274-75. We urge readers who thirst for a more complete understanding of the genesis and operation of the decree to consult that opinion. For present purposes, we are content to note that the Beecher decree was intended to rectify a situation in which many fire departments within Massachusetts had remained lily-white, or nearly so, despite dramatic increases in the African-American and Hispanic populations. The decree sought to accomplish its goal by affirmative action, i.e., by fostering a hiring regime that accorded race-based preferences to blacks and Spanish-surnamed individuals.1 Id. at 1274 n. 2. Each of the affected fire departments was to remain subject to the strictures of the decree (and, thus, to accord race-based preferences) until such time as that department met a general benchmark established by the decree: the attainment of parity (or, at least, rough parity). See Beecher I, 371 F.Supp. at 523 (providing for release from the decree "[a]s a city or town achieves a complement of minorities commensurate with the percentage of minorities within the community"). The meaning of this criterion, and the manner in which it is to be gauged, are questions that have permeated this litigation from the outset. Not surprisingly, those questions are central to the case at hand. — 5 — Unlike some forty-five other fire departments that heretofore have met the goals of the decree and gained release from its constraints, the Boston Fire Department (BFD) has operated under the auspices of the Beecher decree since 1974. A decade ago, a group of non-minority men, aspiring to appointments as firefighters within the BFD, endeavored to bring the City out from under the umbrella of the decree. See Mackin, 969 F.2d at 1275. Although we affirmed the district court's rejection of that attempt, we noted that the decree was not meant to operate in perpetuity. To the contrary, it would remain in force as to any particular municipality only until its stated goal had been achieved. Id. at 1278. A decade has passed, but — despite increased diversity within the BFD — the City still hires firefighters in accordance with the Beecher decree. — 6 — That brings us to the events giving rise to the instant case. In 1998, the Candidates — Joseph Quinn, Sean O'Brien, Robert Dillon, Joseph Sullivan, and Roger Kendrick, Jr. — aspired to firefighter appointments in the BFD. All five are white males; as was required, they identified themselves as non-minority applicants. Each took the firefighter entrance examination administered by the Massachusetts Division of Personnel Administration (MDPA) and scored ninety-nine out of a possible one hundred points. — 7 — The Candidates' scores satisfied the threshold criterion for employment. Along with all other qualifying applicants, they were placed on a civil service eligibility list in rank order. This ranking made allowance for various statutory preferences (e.g., veterans, residents, children of firefighters killed or disabled in the line of duty), ceding pride of place to the holders of such preferences in accordance with state law (even if those persons had earned lower test scores than other qualified candidates). See Mass. Gen. Laws ch. 31, §§ 26, 40. None of these statutory preferences involve race or ethnicity, and none of them are challenged in this proceeding. — 8 — Preparing to hire fifty new firefighters, the BFD requested a certified list of eligible applicants from the Massachusetts Human Resources Division (HRD). The HRD selected individuals in rank order (based on statutory preferences and test scores) and grouped them into a putative "hiring class." After screening out those individuals who stumbled over a variety of race-neutral preconditions (such as drug tests and physical examinations), the HRD composed a slate of "hiring pairs" by placing the highest ranking minority member and the highest ranking non-minority member into a group of two and then repeating the process until the hiring class had been exhausted. — 9 — In November 2000, the BFD, following this rank order, chose twenty-five pairs from the eligibility list and appointed those fifty individuals as entry-level firefighters.2 The Candidates were not among those selected. Although each received a letter from the BFD stating that the vacancies had been filled by persons who outranked him on the certification list, the record suggests that, had the BFD followed a strict rank-order selection process (without any consideration of race or ethnicity), the Candidates (or some of them) likely would have been in the top fifty. — 10 — The Candidates (other than Kendrick, who joined the litigation at a later date) then sued the City. They argued that the City impermissibly had used preferences based on race and ethnicity to rank minorities ahead of them on the eligibility list. In the Candidates' view, this constituted discrimination in violation of the Fourteenth Amendment to the United States Constitution and 42 U.S.C. § 1983 (count 1), Mass. Gen. Laws ch. 151B, § 4 (count 2), and 42 U.S.C. § 2000e et seq. (count 3). The Candidates also attacked another aspect of the hiring process; they argued that requiring them to submit to medical examinations without a conditional offer of employment violated both federal and state law (counts 4-6).3 The district court granted the motion of the Boston Chapter of the NAACP to intervene as a party defendant. See Fed.R.Civ.P. 24. — 11 — The Candidates moved for summary judgment on the first three counts of their complaint or, in the alternative, for a preliminary injunction forbidding the City from filling at least five firefighter positions pending a resolution of the action. They maintained, among other things, that the City should not have applied the Beecher decree to the November 2000 hiring cycle because, by that time, the City had achieved parity in the firefighter force (and, therefore, had met the benchmark for release from the strictures of the decree). The defendants did not controvert the material facts, but, rather, opposed this motion as a matter of law and cross-moved for summary judgment. The district court denied the Candidates' motion and granted summary judgment for the defendants on counts 1 through 3. This appeal followed. Counts 4 through 6 remain pending before the district court. II. APPELLATE JURISDICTION — 12 — It is too familiar a proposition to require citation of authority that a federal court may not act beyond the scope of its jurisdiction. As a logical corollary, parties cannot confer subject matter jurisdiction on a federal court by waiver or consent. See Prou v. United States, 199 F.3d 37, 42 (1st Cir.1999); United States v. Horn, 29 F.3d 754, 767-68 (1st Cir.1994). Consequently, when a court senses a potential lack of subject matter jurisdiction, it ought to inquire further regardless of whether the parties have raised the issue. BIW Deceived v. Local S6, 132 F.3d 824, 828 (1st Cir.1997). We must conduct such an inquiry here. — 13 — In the usual case, an appeal must await the entry of a final judgment, commonly regarded as a judgment that fully disposes of all claims asserted in the action. See Curtiss-Wright Corp. v. Gen. Elec. Co., 446 U.S. 1, 8, 100 S.Ct. 1460, 64 L.Ed.2d 1 (1980); Spiegel v. Trustees of Tufts Coll., 843 F.2d 38, 42 (1st Cir.1988); see also 28 U.S.C. § 1291. There are, however, exceptions to the classic final judgment rule. One such exception, embodied in Rule 54(b) of the Federal Rules of Civil Procedure, allows the immediate entry of a partial final judgment as to fewer than all the claims in a multi-claim action "upon an express determination that there is no just reason for delay." The district court made such a determination here and directed that a separate and final judgment enter as to counts 1 through 3. — 14 — Despite this explicit direction, a jurisdictional problem looms. The law is firmly established in this circuit that a rote recital of Rule 54(b)'s talismanic phrase is not enough, in and of itself, to trump the wonted application of the final judgment rule. See Spiegel, 843 F.2d at 42 (endorsing the "long-settled and prudential policy against the scattershot disposition of litigation"). To warrant recourse to the special procedure envisioned by Rule 54(b), the district court typically must make an individualized assessment of the desirability and effect of an immediate appeal. Id. at 42-43. Thus, if a district court wishes to enter a partial final judgment on the ground that there is no just reason for delay, it should not only make that explicit determination but should also make specific findings and set forth its reasoning. See id. — 15 — We have warned that the parties have an obligation to bring this requirement to the district court's attention. See id. at 44 n. 5. In this instance, the parties did not fulfill this obligation, and the court neither made the requisite findings nor explicated the reasons underlying its Rule 54(b) certification. — 16 — Although this deviation from preferred practice is troubling, it is not necessarily fatal. We have noted that there are "infrequent instances" in which the record, on its face, makes it sufficiently apparent that the circumstances support an appeal from a partial judgment. Id. at 43 n. 4. This is such a case. — 17 — Counts 1 through 3 of the complaint deal with the obligations and protections afforded by the Beecher decree. In contrast, counts 4 through 6 have very little to do with the decree (and nothing to do with race or ethnicity). Moreover, one of the principal parties — the NAACP — has no knowledge about (and, for aught that appears, no interest in) the claims still pending in the lower court. Given the discrete nature of the two sets of claims, the summary judgment decision appears to be "final in the sense that it is an ultimate disposition of ... individual claim[s] entered in the course of a multiple claims action." Curtiss-Wright, 446 U.S. at 7, 100 S.Ct. 1460 (internal quotation marks omitted). — 18 — Equally as important, the proof needed to establish the allegations of counts 1 through 3 is materially different from the proof needed to establish the allegations of counts 4 through 6. By the same token, the legal issues are separate and distinct. Moreover, if the Candidates prevail on one or more of the first three counts and obtain satisfactory relief, counts 4 through 6 may well be rendered moot. This concatenation of circumstances means that, in all probability, there will be no significant duplication of effort in litigating one set of claims to a conclusion and then addressing the remaining set of claims. Such a lack of overlap strongly supports the finding of no just reason for delay (and, thus, the entry of a partial final judgment under Rule 54(b)). See, e.g., Maldonado-Denis v. Castillo-Rodriguez, 23 F.3d 576, 580 (1st Cir.1994); Feinstein v. Resolution Trust Corp., 942 F.2d 34, 40 (1st Cir.1991). — 19 — The most important factor counseling in favor of allowing an immediate appeal in this case is the public interest. As a practical matter, a final resolution of the issues raised in counts 1 through 3 will have a broad impact on the future of all applicants for firefighter positions in the City of Boston. This is a vital concern, as hiring is ongoing. The district court recognized this consideration when it agreed to render an expedited decision on counts 1 through 3. See Quinn, 204 F.Supp.2d at 157 n. 3. In short, the nature of the issue calls out for immediate resolution: time is of the essence if for no other reason than that race-based hiring preferences inevitably shift some of the burden of remediation to innocent persons, Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 280-81, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986) (plurality op.), and thus should not remain in place for any longer than necessary to alleviate the effects of past discrimination. See Regents of the Univ. of Cal. v. Bakke, 438 U.S. 265, 308, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978) (plurality op.) (cautioning that such "remedial action [must] ... work the least harm possible"). — 20 — To sum up, the "findings" requirement that we have superimposed on Rule 54(b) is important, but it is not to be applied woodenly. When the record and the interests of justice permit, we have on occasion relaxed the requirement. See, e.g., Maldonado-Denis, 23 F.3d at 580-81; Feinstein, 942 F.2d at 40. Given the factors enumerated above — especially the substantial public interest that attaches to an expeditious resolution of whether the Beecher decree should continue to constrain the BFD's hiring practices — we conclude that this is one of the rare cases in which, despite the absence of detailed district court findings, the conditions for the use of Rule 54(b) effectively have been met. Accordingly, we have jurisdiction over the Candidates' appeal. III. ANALYSIS — 21 — Having assumed jurisdiction, we turn to the merits of the appeal. We divide our discussion into five segments. First, we frame the issue. We next describe how a court is to assess whether the release point of the Beecher decree — parity — has been achieved, outlining the required comparison and delineating how to arrive at its component parts. Then, we do the mathematics. Finally, we address the question of relief. A. Framing the Issue. — 22 — Parity (or, at least, rough parity) is the key that unlocks the restrictions of the Beecher decree as to a particular community. The central issue in this appeal involves how parity should be calculated for that purpose. The parties agree that the relevant time frame is November of 2000. They also agree as to what sources of data are appropriate for use in this context; all parties accept the City's breakdown of the BFD's complement as of the relevant dates and cite data derived from the 2000 census to establish the appropriate labor pool. Their dispute revolves around what categories of this data govern the measurement of parity. The controlling algorithm has two components. One (which we shall call the first variable) relates to how one measures minority penetration within a particular fire department. The other (which we shall call the second variable) relates to how one measures minority representation in the community as a whole. — 23 — As to these points, the Candidates argue that, under the Beecher decree, the first variable should be limited to the percentage of firefighters in the BFD (not including officers). This should be compared to the second variable, which they envision as the percentage of minorities in the City's employment-eligible population. On the other hand, the City and the intervenor contend that the first variable should represent the percentage of minorities in the BFD as a whole (including officers) and that the second variable should comprise the percentage of minorities in Boston's overall population. — 24 — The district court concluded that the doctrine of stare decisis dictated the composition of both variables. In the court's view, that doctrine required measuring parity by comparing the percentage of minorities in all ranks within the BFD to the percentage of minorities in the City's overall population. Quinn, 204 F.Supp.2d at 161 (accusing the Candidates of "us[ing] a different statistical method than the one used in the Beecher decree"). Since this calculation did not show that parity had been attained — the percentage of minorities within the BFD, including all ranks, was 31.5%, whereas the percentage of minorities in the overall population was slightly over 38% — the court granted the City's motion for summary judgment. Id. at 163. — 25 — We address the nature of the two variables in the next two sections of this opinion. Before doing so, however, we pause to mention two sets of legal principles. First, because this litigation challenges a judicial decree affording race-based relief, any interpretation of the decree or any application of it in practice must survive strict scrutiny. See Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 235, 115 S.Ct. 2097, 132 L.Ed.2d 158 (1995) ("Federal racial classifications, like those of a State, must serve a compelling governmental interest, and must be narrowly tailored to further that interest."); Stuart v. Roache, 951 F.2d 446, 449 (1st Cir.1991) (concluding that all race-based classifications must pass strict scrutiny). This standard obliges an inquiring court to make a binary finding that the race-based relief is not only justified by a compelling governmental purpose but also narrowly drawn to fit the contours of that purpose. Mackin, 969 F.2d at 1275; Stuart, 951 F.2d at 449. — 26 — Second, the district court ruled at the summary judgment stage. The role of summary judgment is to pierce the boilerplate of the pleadings and provide a means for prompt disposition of cases in which no trial-worthy issue exists. See Suarez v. Pueblo Int'l, Inc., 229 F.3d 49, 53 (1st Cir.2000). This device should be employed only when the "pleadings, depositions, answers to interrogatories, and admissions on file, together with the affidavits, if any, show that there is no genuine issue as to any material fact and that the moving party is entitled to a judgment as a matter of law." Fed.R.Civ.P. 56(c). We review an order granting summary judgment de novo. Suarez, 229 F.3d at 53. This non-deferential review is particularly appropriate where, as here, the district court's grant of summary judgment is based on its application of abstract principles of law to essentially uncontradicted facts. See Plumley v. S. Container, Inc., 303 F.3d 364, 369 (1st Cir.2002). B. The First Variable. — 27 — The district court concluded that the first variable to be used for measuring parity under the Beecher decree was the percentage of minority firefighters in the BFD as a whole. The court did not reach this conclusion through independent analysis, but, rather, premised it on the assumption that this issue had been fully litigated and settled in earlier proceedings. See Quinn, 204 F.Supp.2d at 162. Although we appreciate the district court's sensitivity to the role of precedent, an examination of the prior decisions touching upon the Beecher decree refutes the court's assumption. — 28 — To the extent that the Beecher I court compared relative racial representation using quantitative data, it did so only to determine proper placement of the burden of proof as to what discriminatory effects could be attributed to the defendants' preexisting recruitment and hiring practices. See 371 F.Supp. at 514, 518-20. Thus, Beecher I did not mandate a particular parity-measuring formula for future use. That omission is understandable: the minuscule number of minorities in the affected fire departments at the time of the original Beecher decree negated any need for precise definition of how to measure anticipated minority penetration.4 — 29 — This court's initial affirmance of the Beecher decree is unhelpful on this point. In resolving that appeal, we stated that the Beecher decree would "remain[ ] in force, for each local fire department, until that department attains sufficient minority fire fighters to have a percentage on the force approximately equal to the percentage of minorities in the locality," Beecher II, 504 F.2d at 1027. This statement begs the question of who is a "firefighter." — 30 — Our subsequent opinion in Mackin likewise left open the issue of how to measure parity. That case rejected an argument that the second variable was the racial composition of the City as it existed in 1974 (when the Beecher decree was first entered). Mackin, 969 F.2d at 1276. We disposed of that argument without ruling on the significance of the figures presented vis-à-vis the composition of the first variable. See id. Elsewhere in that opinion, we addressed an overbreadth challenge and held that the decree continued to survive strict scrutiny. Id. at 1278. In the process, we cited the language of the decree itself as an indicium of its limited life. See id. (acknowledging that the decree would "remain[ ] in force only until its requirements have been met"). Seen in this light, any elliptical references in Mackin as to how parity might be measured cannot be deemed a holding. — 31 — Our determination that the district court erred in concluding that stare decisis supplies an answer to the question of how the first variable should be constructed brings us back to square one. While we recognize that "district courts enforcing public law consent decrees have, in general, broad discretion in determining such matters as whether the objectives of the decree have been substantially achieved," Navarro-Ayala v. Hernandez-Colon, 951 F.2d 1325, 1337 (1st Cir.1991), the court below did not exercise any such discretion. Instead, it offered its interpretation of judicial precedents dealing with the Beecher decree. Having found that interpretation wanting, we owe no particular deference to the district court's conclusion. See Wessmann v. Gittens, 160 F.3d 790, 795 (1st Cir.1998). — 32 — If a district court's adjudication of a public sector consent decree does not warrant deference, "[o]rdinary contract principles apply" to the same extent as they apply to other consent decrees. Navarro-Ayala, 951 F.2d at 1339. This is especially true when, as now, the question involves determining the parties' original intent and, concomitantly, the scope of the arrangement to which they initially consented.5 Id. "Parties to a consent decree are entitled to know that ... [it] will not be treated as a mere entering wedge which ... gives a district court untrammeled discretion to increase" the depth and breadth of judicial supervision. Id. at 1339 n. 17. — 33 — Against this backdrop, we turn to the parties' competing interpretations of the term "firefighter." As said, the Candidates view that term as encompassing only the lowest rank, that is, entry-level members of the BFD.6 In contradistinction, the defendants view the term as encompassing all uniformed BFD personnel, including officers. In the abstract, it may seem possible to stretch the term "firefighter" to fit the defendants' vision — but this case does not deal with abstractions. Terms in a consent decree cannot be construed in a vacuum; they must instead be read in the context of the decree as a whole. See Newport Plaza Assocs. v. Durfee Attleboro Bank, 985 F.2d 640, 646 (1st Cir.1993); see generally 5 Arthur L. Corbin, Corbin on Contracts § 24.21, at 204 (Joseph M. Perillo rev. ed.1998). From that coign of vantage, the Candidates' definition is much more compelling. — 34 — The litigation underlying the Beecher decree arose out of discriminatory practices in recruitment and hiring into entry-level positions. See Beecher I, 371 F.Supp. at 509 (describing the focal point of the case as the municipalities' "overall hiring policies for the position of firefighter"). Consistent with this focus, the district court made specific findings of discriminatory practices in recruitment and hiring, and the decree was tailored to counteract the effects of that discrimination. Id. at 520. This historical perspective sheds considerable light on the meaning of the term "firefighter" as used in the Beecher decree. Logically, the parties must have intended that term to refer to the class of jobs open to external hiring: entry-level firefighter positions. See Fleet Nat'l Bank v. H & D Entm't, Inc., 96 F.3d 532, 539 (1st Cir.1996) (interpreting terms of a settlement agreement in light of the aim of the litigants at the time of settlement); see generally 5 Corbin, supra § 24.20, at 193. Only by interpreting the term "firefighter" in that manner is it possible to harmonize the wording of the decree with its evident purpose. — 35 — Other language in the decree supports this reading. The decree's first paragraph zeroes in on entry-level positions: it enjoins activities that "discriminat[e] against any applicant or potential applicant for employment." Beecher I, 371 F.Supp. at 521. The decree purports to regulate the recruiting, certifying, and appointing of entry-level personnel and specifically refers to this three-part process as the "hiring procedure." See id. at 522. Subsequent amendments, such as the interim consent decrees entered on April 17, 1975, and November 25, 1975, continue this emphasis on recruitment and hiring for entry-level positions. — 36 — Custom and usage within an affected industry or workplace can be important aids to the construction of a contract or consent decree. See, e.g., Boston Police Superior Officers Fed'n v. City of Boston, 147 F.3d 13, 17-18 (1st Cir.1998) (interpreting terms with otherwise broad usage narrowly according to usage within the relevant profession); Smart v. Gillette Co. Long-Term Disab. Plan, 70 F.3d 173, 179 (1st Cir.1995) (noting that "usages of trade" are relevant to interpreting terms of a contract). Here, the usage of the BFD supports the Candidates' definition of the term "firefighter." — 37 — In the court below, the evidence showing the composition of the BFD differentiated the various ranks by specific title. These records consistently referred to the entry-level rank as "Firefighter" (sometimes written as "Fire Fighter" or "Fire fighter"). Corresponding to that term, the evidence showed the number of persons serving in the entry-level rank, broken down by race and ethnicity. That evidence referred to the other ranks — the officers — by such titles as "Fire Lieutenant," "Fire Captain," and "District Fire Chief," giving the racial and ethnic composition of each such rank. This indicates that, within the BFD, the term "firefighter" carries with it specific duties, responsibilities, pay, and privileges. — 38 — Lexigraphic insights are frequently helpful in determining the meaning of specific words in consent decrees. See, e.g., Gilday v. Dubois, 124 F.3d 277, 285-86 (1st Cir.1997). Here, the mine-run of dictionary definitions tends to support the Candidates' version of the disputed term. A firefighter is said to be "one who fights fires as a member of a municipal fire department," Webster's Third New Int'l Dictionary 855 (1993) (internal dictionary symbols omitted), or "a person who fights destructive fires," Random House Dictionary of the English Language 722 (2d ed.1987), or, simply, "a person who fights fires," Merriam-Webster On-Line Dictionary (2002), at http://www.mw.com/home.htm. These definitions uniformly emphasize the principal activity that firefighters perform. They show quite clearly that the term "firefighter," in its ordinary sense, applies more naturally to those whose exclusive province is attempting to extinguish conflagrations as opposed to those in the higher echelons (who must handle a broader and more diversified range of activities, including administration and supervision). While the dictionary definitions of "firefighter" do not necessarily exclude those whom the defendants seek to include, they argue persuasively against their inclusion. — 39 — Our case law also militates against including higher ranks under the appellation "firefighter." For example, in Boston Police Superior Officers, 147 F.3d at 17-18, we held that the term "police officers" used in a consent decree governing promotions within the Boston Police Department did not include all police officers, but, rather, encompassed only patrolmen. That case, like this one, illustrates the importance of honoring contextual constraints that may be placed on a word or phrase within the four corners of a consent decree. — 40 — Last — but far from least — including higher ranks within the compass of the term "firefighter" would mean that the Beecher decree was not narrowly tailored to serve its stated purpose. Such an interpretation would effectively transform an instrument carefully crafted to eliminate discrimination in recruitment and hiring into general leverage favoring minorities in a much wider variety of matters. In most of those areas — promotion is a good example — no justification for systematic preferential treatment of minorities has been established. Thus, the transformation that necessarily would accompany an acceptance of the defendants' interpretive gloss would present an insurmountable constitutional obstacle. See Bakke, 438 U.S. at 307, 98 S.Ct. 2733. Because courts should avoid construing a consent decree in a constitutionally offensive way if a feasible alternative construction exists, this looming constitutional quandary is itself a cogent argument for rejecting the defendants' expansive view of the term "firefighter." Cf. Walsh v. Schlecht, 429 U.S. 401, 408, 97 S.Ct. 679, 50 L.Ed.2d 641 (1977) ("[C]ontracts should not be interpreted to render them illegal and unenforceable where the wording lends itself to a logically acceptable construction that renders them legal and enforceable."). — 41 — The opposite pan of the scale contains nothing of sufficient weight to counterbalance this powerful asseverational array. Insofar as we can discern, interpreting the word "firefighter" to include higher ranks has no basis in the record. The decree is utterly silent as to activities affecting higher ranks, and the remainder of the record — both current and historical — is devoid of any allegation or finding that the BFD has ever discriminated against minorities in promotions or in other personnel practices involving those in (or aspiring to reach) the higher ranks. Indeed, the original Beecher plaintiffs would not have had standing to allege discrimination in such respects.7 Taking these facts into account, it unfairly belittles the Beecher court to read "firefighter" as that term is used in the decree to include all ranks (and, thus, to impute to the court the injudicious crafting of a remedy that reaches beyond the scope of the surrounding litigation). — 42 — The defendants, ably represented, nonetheless advance a plethora of counter-arguments. Only four of them require discussion. — 43 — First, the City and the intervenor point to paragraph 11 of the interim consent decree of November 25, 1975. That proviso requires a municipality seeking to be released from the decree to petition the MDPA, informing that agency "that the percentage of post-probationary minority uniformed personnel equals the percentage of minorities" in that municipality. We do not believe that this casual (and wholly unexplained) linguistic switch from "firefighter" to "uniformed personnel" can be accorded decretory significance. This is especially so in light of two facts. First, that same proviso incorporates by reference paragraph 7 of the April 17, 1975 interim consent decree (which opens the way for release from the decree "whenever a particular city or town has achieved a complement of Black or Spanish-surnamed firefighters commensurate with the percentage of minorities within the community"). Second, the preamble to the November 25 decree — the decree which introduced the term "uniformed personnel" into the litigation — unequivocally declared that "the parties consent to entry of this Agreement to permit ... the establishment of an eligibility list for the position of firefighter ...." (Emphasis supplied). In the final analysis, then, the phrase upon which the defendants rely leads us back to our starting point. — 44 — We add, moreover, that if the phraseology employed in the November 25, 1975 decree was intended to clarify that the earlier decrees affected more than the firefighter rank, it seems highly likely that the Beecher court would have used more precise language such as "of all ranks" or "of any rank." But the Beecher court eschewed such language. What is more, the court demonstrated an inclination to use general descriptors for specific subject matter. See Beecher I, 371 F.Supp. at 523 (using the general phrase "complement of minorities" to refer to the specific minority "[g]roup of all eligible blacks and Spanish-surnamed persons"). In our view, this proclivity undermines the inference that the defendants (and the dissent) seek to draw from the phrase "post-probationary uniformed personnel." — 45 — Before leaving this point, we note that our dissenting colleague sees "no basis for concluding that `firefighter' is the operative word" in the Beecher decree and the two interim decrees that followed. Yet that word appears no fewer than twenty-six times in those decrees, while the term "uniformed personnel" appears only once. This repeated emphasis, especially when coupled with the fact that the litigation revolved around the position of firefighter and did not concern other ranks, makes it perfectly clear that "firefighter" is indeed the operative term. — 46 — Next, the defendants strive to convince us that we should defer to a 1987 memorandum issued by the MDPA, which indicates that, for Beecher purposes, the first variable should consist of "the actual percentage of the authorized uniformed permanent full-time force which is made up of tenured Black or Hispanic officers of any rank." We are not persuaded. — 47 — The deference typically owed by a court to an administrative agency derives from the fact that the agency has been entrusted by a legislative body to administer a statute enacted under that branch's separate constitutional authority. See Chevron, U.S.A. Inc. v. Natural Res. Def. Council, Inc., 467 U.S. 837, 842-44, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984); see also Fireside Nissan, Inc. v. Fanning, 30 F.3d 206, 212 (1st Cir.1994) (reasoning that state officials' interpretation of state statute is entitled to deference). If the legislative branch has "left a gap for the agency to fill, there is an express delegation of authority." Chevron, 467 U.S. at 843-44, 104 S.Ct. 2778. Thus, a court must defer to the agency's reasonable construction of a provision in the statute. Id. at 844, 104 S.Ct. 2778. — 48 — Here, however, a state agency is helping to administer compliance not with a legislative enactment but with a federal judicial decree. A federal appellate court owes no deference to such an agency when endeavoring to discern the meaning and constitutional limits of such a decree. After all, if the judiciary is the final arbiter as to questions of statutory construction and must refuse to accept administrative interpretations that contradict clearly ascertainable legislative intent, see id. at 843 n. 9, 104 S.Ct. 2778, courts certainly have the power — indeed, the duty — to reject an administrative construction that runs contrary to the manifest intent of a judicial decree. — 49 — Third, the defendants posit a variation on the "trickle up" theory. See, e.g., Stuart, 951 F.2d at 452 ("One obvious reason ... why there may have been few black sergeants in the Boston Police Force in 1978 is that the Department had not hired many black police officers before 1970. Since unjustified discrimination accounted for the latter fact, [it] cannot excuse the former."). They assert that not including higher ranks within the term "firefighter" would diminish the efficacy of the decree and reward appointing authorities for keeping minorities in the firefighter rank (thus perpetuating underrepresentation in the higher echelons). For that reason, they argue, it makes sense to require parity department-wide before allowing a community to escape from the constraints imposed by the Beecher decree. — 50 — This argument is specious. We have determined, and the defendants concede, that the parties never intended the decree to address promotions. On that basis alone, it would be improper to extend the decree's reach by judicial fiat. See Boston Police Superior Officers, 147 F.3d at 17-18. Furthermore, alleging that unconstitutional hiring practices within the BFD have caused a disparate impact on the number of minorities in the higher echelons is a heavy charge. Such an allegation should not be addressed in the abstract, but, rather, should be squarely propounded by an injured party with standing to sue and litigated to a just conclusion. Because the litigation giving rise to the Beecher decree is of a different nature than the justification that the defendants now offer, the decree is a constitutionally insufficient vehicle for addressing that justification. See id. at 18; see also Wessmann, 160 F.3d at 802 ("The mere fact that an institution once was found to have practiced discrimination is insufficient, in and of itself, to satisfy a state actor's burden of producing the reliable evidence required to uphold race-based action."). — 51 — Finally, the City and the intervenor make an argument of last resort. They contend that their interpretation of the term "firefighter" is proper because they have assumed all along that the term embraced more than entry-level personnel. In the circumstances of this case, such a contention does not get them very far. — 52 — In pressing this point, the defendants lean heavily upon past practice of the MDPA and attempt to apply language derived from Mackin to that practice. There, we rejected the argument that static 1974 population figures controlled the measurement of parity. In so doing, we wrote that "[f]ew things evidence a decree's meaning more persuasively than an immutable, decade-old pattern of past practice under the decree, consensually engaged in by all sides in the underlying litigation that produced the decree." 969 F.2d at 1276. — 53 — That language is inapposite here. In Mackin, we were using a uniformly accepted past practice to validate a facially plausible interpretation of the Beecher decree (i.e., that current population figures would control). See id. We deemed that past practice confirmatory of what appeared to be a commonsense reading of the verbiage chosen by the parties and approved by the district court. In contrast, the defendants here attempt to use an administrative practice first inaugurated in 1987 (thirteen years after the entry of the Beecher decree), and never directly challenged, to change the meaning of plain language and thus to validate an unlikely interpretation that is implausible in the context of the underlying litigation (and one which, in the bargain, would open the decree to constitutional attack). To the extent that the MDPA's past practice points toward that suggested interpretation, it lacks record support and therefore cannot dominate the decisional calculus. See Wessmann, 160 F.3d at 802; Stuart, 951 F.2d at 454-55. — 54 — To dwell on this issue would be pointless. It suffices to say that the history and purpose of the litigation, the language of the decree, and the usage of the BFD make manifest that the term "firefighter," as used in the Beecher decree, was intended to include only those persons serving in the entry-level rank. Interpreting "firefighter" more broadly would rip the decree from its constitutional moorings, twist its language into something alien to its evident purpose, and violate settled principles of contract law. Accordingly, the proper parity-measuring device — what we have called the first variable — comprises the percentage of minorities in the entry-level rank of the BFD as of the relevant date (November of 2000). C. The Second Variable. — 55 — As for the comparison that Beecher demands — the second variable — the Candidates, citing cases decided after the entry of the original Beecher decree, assert that in affirmative action employment cases minority representation must be measured against the specific adult population that comprises the qualified labor pool. See City of Richmond v. J.A. Croson Co., 488 U.S. 469, 501-02, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989); Local 28 of Sheet Metal Workers' Int'l Ass'n v. EEOC, 478 U.S. 421, 479, 106 S.Ct. 3019, 92 L.Ed.2d 344 (1986); Stuart, 951 F.2d at 450, 453-54. They insist that any other focus would be unconstitutional and that, therefore, the relevant comparison in this case necessarily must be to Boston's over-19 black and Hispanic population. See Mass. Gen. Laws ch. 31, § 58 (requiring firefighters to be over 19 years of age).8 — 56 — The district court held that this point previously had been resolved. Quinn, 204 F.Supp.2d at 163. We agree that stare decisis governs. The last time around, we addressed this aspect of the standard for measuring compliance head-on and held that the proper data are the "contemporaneous population figures [for] ... `the percentage of minorities within the community.'" Mackin, 969 F.2d at 1276 (quoting the Beecher decree). At that time, we considered the very authorities upon which the Candidates now rely and specifically repudiated the notion that the pertinent language of the decree was overbroad. Id. at 1277-78. Thus, principles of stare decisis required the district court to reject the Candidates' version of how the second variable should be constructed. See Planned Parenthood v. Casey, 505 U.S. 833, 854, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992) (recognizing that the doctrine of stare decisis embodies "[t]he obligation to follow precedent"). — 57 — If more were needed — and we do not think that it is — we note that the Beecher decree unambiguously requires the use of the percentage of minorities in the general population as the second variable for gauging discriminatory patterns in entry-level hiring. See Beecher I, 371 F.Supp. at 523; see also Mackin, 969 F.2d at 1276 (enumerating portions of the Beecher decree that require, or plainly contemplate, reference to the percentage of minorities in the general population). That requirement is fully consistent with the authorities cited by the Candidates. See Croson, 488 U.S. at 501, 109 S.Ct. 706 ("In the employment context, we have recognized that for certain entry level positions requiring minimal training, statistical comparisons of the racial composition of the relevant population may be probative of a pattern of discrimination."). Whether or not some other standard might be more precise, the fact remains that the firefighter position is of the type contemplated by the Croson Court. Thus, the parties' original bargain for a community-wide pool and the district court's endorsement of that concept are entitled to great weight.9 See Navarro-Ayala, 951 F.2d at 1339 (reasoning that the district court is best able to decide certain types of parameters in public institutional reform litigation and, thus, is owed deference when the scope of the parties' original bargain is not at issue). — 58 — We reaffirm, therefore, that the appropriate standard for measuring compliance with the Beecher decree — what we have termed the second variable — consists of contemporaneous population statistics depicting the percentage of minorities within the overall community. It follows that a fire department subject to the Beecher decree will remain so until that department succeeds in demonstrating that it has achieved "the decree's actual objective: rough parity," Mackin, 969 F.2d at 1277, measured by establishing the percentage of minorities among entry-level firefighters in the particular department and comparing that ratio to the percentage of minorities within the general population of the local community. When this comparison demonstrates that parity (or, at least, rough parity) has been attained, then the Beecher decree has outlived its usefulness as to that community's firefighting force. D. Reworking the Algorithm. — 59 — From this analysis, it is evident that the district court erred in constructing the first Beecher variable. In order to determine whether that error affected the Candidates' substantial rights, see Fed. R.Civ.P. 61, we rework the algorithm using the proper variables. If that algorithm does not yield ratios that show parity (or, at least, rough parity) between the percentage of minorities in the entry-level rank of the BFD and the percentage of minorities in the City of Boston as a whole, then the entry of summary judgment in the City's favor must stand. See Houlton Citizens' Coalition v. Town of Houlton, 175 F.3d 178, 184 (1st Cir.1999) (holding that a grant of summary judgment may be affirmed on any independent ground revealed by the record). — 60 — On this issue, the parties urging that parity has been achieved (here, the Candidates) bear the burden of proof. See C.K. Smith & Co., Inc. v. Motiva Enter. LLC, 269 F.3d 70, 73 (1st Cir.2001). That is of purely academic interest in this case for the pertinent facts are uncontradicted. The court below developed the factual record sufficiently to demonstrate that, when the City recruited the 2000 hiring class, blacks and Hispanics comprised approximately 40% of the firefighters within the BFD. At the same time, blacks and Hispanics constituted slightly over 38% of Boston's overall population. Quinn, 204 F.Supp.2d at 162. Hence, parity had been achieved, and the City had become eligible for release from the strictures of the Beecher decree. See Beecher II, 504 F.2d at 1026-27. — 61 — Given these facts, the district court's error was not harmless. After all, a public employer who consents to the use of race as a factor in order to palliate the lingering effects of past discrimination must maintain continuous oversight in order to ensure that the decree works the least possible harm to other innocent persons competing for employment. Bakke, 438 U.S. at 308, 98 S.Ct. 2733. Once parity has been achieved, the decree has served its legitimate purpose, and the justification for it has abated. See id. at 309, 98 S.Ct. 2733. From that point forward, the employer has no basis to continue preferring minorities. See id.; see also Mackin, 969 F.2d at 1276 ("An intrusion by a federal court into the affairs of local government should be kept to a bare minimum and not be allowed to continue after the violation has abated and its pernicious effects have been cured."). — 62 — We conclude, therefore, that the City's continued resort to race-based preferences from and after the time when parity was achieved fails the second prong of the strict scrutiny analysis. See Bakke, 438 U.S. at 309, 98 S.Ct. 2733. Thus, the City's adherence to the Beecher decree during the 2000 hiring cycle was unconstitutional. Consistent with the foregoing, we reverse the district court's entry of summary judgment in favor of the City and direct the court to enter judgment in the Candidates' favor. E. Relief. — 63 — Our holding does not end the matter: there remains the question of what remedy is appropriate under the circumstances. That is a highly ramified question, the answer to which is subject to the push and pull of competing centrifugal and centripetal forces. — 64 — On the one hand, a public employer's good faith reliance on a consent decree entered to remedy the effects of past discrimination should not lightly be disturbed. See, e.g., Boston Police Superior Officers, 147 F.3d at 25. That proposition has particular bite as to the firefighters who were appointed to the BFD during the 2000 hiring cycle, for the Beecher decree does not directly dictate hiring decisions, but, rather, simply directs municipalities to take affirmative action concerns into account in choosing among qualified applicants. See Mackin, 969 F.2d at 1278 (explaining that "only qualified minority candidates are specially advantaged; no minority candidate is placed on the eligibility list unless he or she has attained a passing score on the entrance examination") (emphasis in the original). Constitutional violations rarely, if ever, justify the reversal of specific employment decisions previously made by government employers insofar as those decisions affect innocent parties. See, e.g., Boston Police Superior Officers, 147 F.3d at 25; cf. Wygant, 476 U.S. at 282-83, 106 S.Ct. 1842 ("Denial of a future employment opportunity is not as intrusive as loss of an existing job."). The history of this case offers a striking example of this truism: the Beecher court, despite its finding of pervasive past discrimination, did not disturb the livelihood of those firefighters who were on the force in 1974. See Beecher I, 371 F.Supp. at 520 (noting the need to "preserv[e] morale within the fire departments"). — 65 — On the other hand, we have acknowledged that when job applicants have been denied equal protection of the laws by a public employer, "[t]he result, not the specific intent, is what matters." Beecher II, 504 F.2d at 1021 (citation and internal quotation marks omitted). In that spirit, the original Beecher court invoked its equitable powers to fashion affirmative relief for a perceived injury even after explicitly disavowing any finding that the "exclusionary policy [giving rise to the litigation] ha[d] been followed intentionally or by design." Beecher I, 371 F.Supp. at 519. — 66 — Here, the Candidates have neither offered any evidence that the City acted in bad faith nor otherwise set forth a plausible basis that would justify a court in second-guessing the specific decisions made by the City in the 2000 hiring cycle. In a similar vein, they have not established that the persons actually hired were either complicit in some scheme or somehow undeserving of their appointments. Consequently, there is no principled basis for turning back the clock and revoking those appointments. — 67 — That does not mean, however, that the Candidates are entitled to no more than a handshake or a tip of the hat. The fact remains that they have succeeded in showing that the City applied a consent decree, previously held to be constitutional, for too long. The result was that the City infringed the Candidates' constitutional rights when it acted upon their applications for appointment to the BFD. They should at least have had the opportunity to compete for the positions that they coveted free of the constraints imposed by the Beecher decree. See Bakke, 438 U.S. at 308, 98 S.Ct. 2733. Balancing these competing considerations, we hold that the BFD's decisions to appoint specific individuals during the 2000 hiring cycle were valid exercises of its responsibility to provide the citizens of Boston with a full complement of qualified firefighters. See Beecher I, 371 F.Supp. at 520. Nevertheless, the application of the decree to the 2000 hiring cycle violated the Candidates' constitutional rights by depriving them of an equal opportunity to compete for positions on the BFD. See Bakke, 438 U.S. at 308, 98 S.Ct. 2733. On remand, the district court must sort through this tangle and determine, in its sound discretion, the appropriate remedy for the Candidates' injury (excluding, however, any form of relief that would require dismissal of any provisionally or permanently appointed firefighter currently serving in the BFD). IV. CONCLUSION — 68 — The goal of the Beecher decree was to eliminate discriminatory practices in the recruitment and hiring of firefighters in communities subject to the Massachusetts Civil Service laws, and, relatedly, to remedy the effects of past discrimination in recruitment and hiring. Remediation has taken more than a quarter-century. At long last, however, that objective has been achieved with respect to the BFD; parity has been reached between the percentage of minority firefighters in the BFD and the percentage of minorities in the City as a whole.10 — 69 — Although this is a significant landmark along the road to equality, we add a word of caution. We are not Pollyannas, and we recognize that achieving parity at the firefighter level is not tantamount to saying that all is well in regard to racial and ethnic issues within the BFD as a whole. To the extent that inequalities remain, however, they are not within the compass of either the Beecher decree or this litigation. Nor will we reach out for them — issues of constitutional magnitude should not be the subject of speculation, but, rather, should be litigated fully by parties with standing to represent various pertinent points of view. For today, we fulfill our responsibility by holding that the City's appointment of firefighters ought no longer be subject to the strictures of the Beecher decree. We need go no further. — 70 — The judgment of the district court is reversed and the case is remanded for further proceedings consistent with this opinion. Costs are to be taxed in favor of the appellants. — 71 — LIPEZ, Circuit Judge, dissenting. — 72 — The majority frames the critical issue as one involving "competing interpretations of the term firefighter." I respectfully disagree that this is the central issue in the dispute. Indeed, the majority attributes an argument to the Candidates that they have not made in this litigation — that the court and the parties to the Beecher decree understood as a matter of original intent that minority presence in the fire departments subject to the decree would be calculated by considering only the entry-level firefighter rank. The Candidates acknowledge that "[w]hen the Beecher decree was originally conceived, its apparent goal was to generate racial parity in the Fire Department as a whole, as opposed to merely the firefighter ranks." Instead, the Candidates argue that we should focus only on the entry-level ranks because changed circumstances since 1974 and 1992, when we last reviewed the Beecher decree in Mackin v. City of Boston, 969 F.2d 1273 (1st Cir.1992), render the original intent of the decree unconstitutional. — 73 — Moreover, even if the Candidates had asked us to determine the original intent of the parties on the calculation of parity, it is a mistake to frame the issue as a search for the meaning of the term "firefighter." As I will explain in more detail, there is no consistent use of the word "firefighter" as the operative term in the calculation of parity. Instead, there are a variety of terms used, only one of which is the word "firefighter." The relevant question, then, is which of the early formulations of the measurement of parity captures the intent of the parties and the court. — 74 — This dissent proceeds in three parts. In part I, I address the majority's conclusion that the Beecher decree always contemplated that the first variable, to use the majority's language, would only consider the percentage of minorities in the entry-level firefighter ranks. In Part II, I address the majority's assertion that narrow tailoring requires this interpretation of the decree to avoid constitutional problems. In Part III, I evaluate the Candidates' argument that changed circumstances since the entry of the Beecher decree require that the decree now be read to measure parity by focusing only on the firefighter ranks. I. Interpreting the Decree — 75 — In the original Beecher decree issued on February 8, 1974, the court used the phrase "complement of minorities" in the operative sentence explaining when a city's fire department could receive an exemption.11 Boston Chapter, NAACP, Inc. v. Beecher, 371 F.Supp. 507, 523 (D.Mass.), aff'd 504 F.2d 1017 (1st Cir.1974). In its Interim Consent Decree of April 17, 1975 (signed by the parties), the court for the first time used the term "firefighter" in explaining the process for obtaining an exemption: "The Director [of Personnel Administration] shall notify the plaintiffs and supply evidence to the plaintiffs whenever a particular city or town has achieved a complement of Black and Spanish-surnamed firefighters commensurate with the percentage of minorities within the community ...." Boston Chapter, NAACP, Inc. v. Beecher, No. 72-3060-F, slip op. at 4 (D.Mass. Apr. 17, 1975) (emphasis added). Soon thereafter, though, the parties signed the "Agreement to Effectuate Interim Consent Decree," which again discusses exemptions. Paragraph 11 of that Agreement states: "No fire department ... may be exempted ... unless the appointing authority has first petitioned the Division that the percentage of post-probationary minority uniformed personnel equals the percentage of minorities in the city or town served by said department ...." Boston Chapter, NAACP, Inc. v. Beecher, No. 72-3060-F, slip op. at 6 (D.Mass. Nov. 25, 1975) (emphasis added).12 — 76 — Presented with these three formulations, the majority chooses to focus on the word "firefighter" in the April 1975 Interim Consent Decree as the critical word and dismisses the formulation in the final agreement of the parties ("the percentage of post-probationary minority uniformed personnel") as a "casual (and wholly unexplained) linguistic switch." This dismissive characterization seems odd, especially since the phrase "post-probationary minority uniformed personnel" in the November 1975 Agreement is far closer to the general language of the original Beecher decree ("complement of minorities") than the word "firefighter" used in the April Interim Decree. The "complement of minorities" and "post-probationary minority uniformed personnel" formulations also seem to capture the purpose that the Beecher court attributed to its own decree — to remedy the "exclusion of minorities from the fire department" — than does the word "firefighter." Beecher, 371 F.Supp. at 519-20. — 77 — However, there is support for competing views in the shifting word choices of these three decrees. The language itself does not tell us definitively what the court and parties intended for the measure of the first variable. But the use of the word "firefighter" is part of the ambiguity — not the answer to it. In the sequence of language in the three decrees, there is no basis for concluding that "firefighter" is the operative word and that determining its plain meaning will answer the intent question.13 Instead, we must acknowledge that the competing formulations of the first variable create an inescapable ambiguity which must be resolved by resort to extrinsic evidence. — 78 — We have already opined in this litigation on the kind of persuasive evidence that helps resolve such ambiguities. "Few things evidence a decree's meaning more persuasively than [a] ... pattern of past practice under the decree ...." Mackin, 969 F.2d at 1276. The City and the NAACP produced evidence that Massachusetts has required municipalities seeking exemptions from the decree to compare the percentage of minorities in the population of their city or town to the percentage of minority personnel among their departments' entire uniformed personnel. Using this formula, forty-five municipalities originally subject to the decree have received exemptions. An affidavit in the record explains that no department's appointing authority has objected to this formula since it has been in place. This understanding of the decree has been memorialized in a 1987 memorandum circulated by the Massachusetts Department of Personnel Administration outlining "the process and the criteria for seeking approval for exemption" from the Beecher decree. This memorandum, using the same terminology as the November 1975 Agreement, states that a municipality must consider the "full-time uniformed force" and calculate the percentage of "post-probationary (tenured) full-time permanent Black or Hispanic members of the uniformed force" to determine whether parity has been achieved. — 79 — The majority dismisses the importance of this memorandum, concluding that "a federal appellate court owes no deference to [a state agency helping to administer compliance with a federal judicial decree] when endeavoring to discern the meaning and constitutional limits of such a decree." As a generality this point is a fair one. But this memorandum from the Massachusetts Department of Personnel Administration demonstrates much more than a reasonable interpretation of ambiguous language in a federal decree by a helpful state agency. This memorandum describes a "pattern of past practice" of the application of language in the November 1975 Agreement by one of the parties to the original litigation. — 80 — The Commissioners of the Massachusetts Division of Civil Service were some of the named defendants in the consolidated Beecher litigation. At some point, the Department of Personnel Administration became the successor to the Division of Civil Services and responsible for carrying out the remedial provisions of the decree. Paragraph 1 of the April 1975 decree describes how the "Director of Civil Service [hereinafter Director shall mean the Director, his successors, or after July 1, 1975, the Personnel Administrator]" should implement the original Beecher decree.14 Paragraph 1 of the November 1975 Agreement refers to "The Division15 of Personnel Administration (formerly known as the Division of Civil Service, and hereinafter, the `Division') ...." Paragraph 11 of the November 1975 Agreement designates the "Division" as the agency to which municipal fire departments should apply for exemptions from the decree. The November 1975 Agreement was also signed by counsel for the "Director of Personnel Administration." The memorandum in question, then, was not issued by an agency that was a stranger to the underlying litigation. Instead, it was crafted by a party to the Beecher litigation with the court-ordered responsibility to review the petitions of municipal fire departments for exemption from the decree. Under these circumstances, the insistence in Mackin on the persuasive value of a pattern of past practice under the decree is particularly apt. See also Navarro-Ayala v. Hernandez-Colon, 951 F.2d 1325, 1353 (1st Cir.1991) (Cyr, J., concurring in part and dissenting in part) (analogizing a consent decree to a contract: "Under a well-established rule of contract interpretation, the court may look to the parties' post-contract course of conduct and performance to ascertain the `practical interpretation and application' that the parties themselves attached to the ambiguous contract language.") (emphasis in original). — 81 — The majority says that Mackin's directive to consider post-execution practice to interpret the decree's meaning is "inapposite" in this case because the evidence of the practice in question is being used "to change the meaning of plain language and thus to validate an unlikely interpretation" of the Beecher decree. However, there is no plain language in the decrees on the issue of the first variable, and the "unlikely interpretation" was one adopted by one of the parties to the underlying litigation and complied with by many other municipalities subject to the decree. Therefore, the majority's interpretation of the decree runs contrary both to Massachusetts's demonstrated understanding of it and to the lengthy history of its application in granting many exemptions from the decree over a period of at least sixteen years (that is, since 1987). — 82 — It is also difficult to reconcile the majority's interpretation of the decree with incentives that would have seemed sensible to the parties to the decree. If, at the time the decree was entered, a city fire department knew that it needed to show parity only in its entry-level class to receive an exemption from the decree, it could achieve its goals more quickly by keeping all the minority firefighters it hired at the entry level — that is, by selecting for promotion only non-minority firefighters. The majority's interpretation would have provided an incentive for fire departments not to promote the black and Hispanic firefighters it hired under the terms of the decree. Every minority firefighter promoted to an officer position would have been another black or Hispanic member of the department that did not count towards the calculation of parity. — 83 — Although there is no evidence in the record that the Boston Fire Department ("BFD") has ever discriminated against black and Hispanic firefighters in its previous promotion decisions, the majority's interpretation of the decree creates an incentive for discriminatory behavior for those municipalities still subject to the decree after the resolution of this case. I cannot imagine that the plaintiffs in the Beecher litigation and the district court would have decided upon a remedy that provides incentives so at odds with the purpose of the litigation and the relief ordered.16 II. Narrow Tailoring — 84 — To support its interpretation of the Beecher decree, the majority contends that the interpretation advanced by the city and the NAACP "would mean that the Beecher decree was not narrowly tailored to serve its stated purpose." In elaborating on this point, the majority raises another argument that was not raised by the Candidates: — 85 — Such an interpretation would effectively transform an instrument carefully crafted to eliminate discrimination in recruitment and hiring into general leverage favoring minorities in a much wider variety of matters. In most of those areas — promotion is a good example — no justification for systematic preferential treatment of minorities has been established. — 86 — As I understand it, the majority reasons that the interpretation of the Beecher decree advocated by the NAACP and the city would impermissibly benefit minority firefighters by somehow giving them preferences for promotions even though the Beecher court made no findings regarding the constitutionality of the promotion policies of the fire departments subject to the decree. This effect of the Beecher decree, the majority concludes, would render it unconstitutional. — 87 — I disagree with this conclusion because the decree does not in fact have the effect of granting minorities preferential treatment in promotions. That is, I fail to see how merely calculating parity by taking the entire department into consideration leads to a "systematic preferential treatment of minorities" in the promotion process. First, there is no evidence in the record that minority firefighters hired in accordance with the Beecher decree have received preferences for promotions. Therefore, the majority's conclusion about such an effect is entirely speculative. Second, the logic of the majority's "preferential treatment" argument is problematic. By its terms, the Beecher decree does not explicitly address the promotion policies of the BFD or any other fire department in any way. Whether parity is calculated by taking into consideration merely entry-level firefighters, or by also including lieutenants, captains, and chiefs, the decree does not mandate changes in the department's promotion practices. Because the criteria for promoting firefighters, whatever they may be, are independent from the hiring criteria, the Beecher court's alterations to the BFD's hiring criteria have no effect on its promotion criteria. Therefore, whether or not the Beecher decree continues to direct the method by which the department hires new firefighters, we can only assume, without any evidence to the contrary, that the department will continue to decide promotions in the same way it always has. — 88 — True, by design, the decree results in an increase in the diversity of the class from which the BFD selects the firefighters for promotion. However, this consequence of the decree cannot be "general leverage favoring minorities." In any organization that promotes from within, the application of a hiring decree that results in the employment of minorities that otherwise would not have been hired necessarily increases the possibility that some of those minorities will be promoted to officers and thereby change the racial makeup of the officer ranks. Just because the decree has the consequence of changing the racial composition of the pool of those eligible for promotions does not mean that it is unconstitutional. III. Changed Circumstances — 89 — As noted, the Candidates concede that, "[w]hen the Beecher decree was originally conceived, its apparent goal was to generate racial parity in the Fire Department as a whole, as opposed to merely the firefighter ranks." However, they argue that after twenty-eight years, "a different analysis must be used if the decree is to be implemented in a constitutionally acceptable manner." Specifically, the Candidates point to three changed circumstances that they claim require a revised interpretation of the decree. First, they say that it has become evident over time that the decree is ineffective for the purpose of generating racial diversity in the upper ranks of the BFD. Second, the Candidates remind us that twenty-eight years have elapsed since the entry of the Beecher decree, and argue that this is simply too long to wait for the realization of rough parity. Third, when the scope of the inquiry is restricted to consideration of only entry-level firefighters, statistics show that the percentage of minority firefighters exceeds the percentage of minorities in the community, resulting in an "excess of parity" at the entry level. — 90 — Frankly, I am at a loss to understand this third "changed circumstance" as a discrete argument. The Candidates offer two variations of "excess of parity." In the first, the Candidates use a computation of parity that the majority has rightly rejected. That computation reflects the difference between the percentage of minority firefighters (39.9%) and the minority population of Boston aged 18 and older (32.6%). I agree with the majority that stare decisis requires the use of the percentage of minorities in the general population, rather than the adult population, as the second variable for gauging parity. In the alternative, the Candidates assert that "even assuming that general population statistics are to be considered, the minority composition of the Department's firefighters (39.9%) exceeded the minority composition of the general population of Boston (38.3%)." However, after conceding that "[w]hen the Beecher decree was originally conceived, its apparent goal was to generate racial parity in the Fire Department as a whole, as opposed to merely the firefighter ranks," the Candidates do not use this formulation of "excess of parity" to articulate a distinct constitutional argument. Instead, they use the statistics as evidence that the decree is not effective for the purposes of achieving this "apparent goal" of racial diversity in the upper ranks of the BFD. Hence, because "excess of parity" in the firefighter rank is inescapably tied to the first "changed circumstance" argument relating to the ineffectiveness of the decree, I will not address this "excess of parity" argument independently. Instead, I will incorporate my discussion of excess parity in the ineffectiveness argument. — 91 — While modification of a consent decree is warranted if there is "a significant change either in factual conditions or in law," Rufo v. Inmates of Suffolk County Jail, 502 U.S. 367, 384, 112 S.Ct. 748, 116 L.Ed.2d 867 (1992), "a party seeking modification of a consent decree bears the burden of establishing that a significant change in circumstances warrants revision of the decree." Id. at 383, 112 S.Ct. 748. The Candidates have failed to carry this burden. — 92 — A. Ineffectiveness of the Decree. — 93 — The Candidates assert that although 39.9% of entry-level firefighters in the BFD are minorities, only 6.2% (22 of 350) of uniformed personnel above the rank of firefighter are minorities. They assert that "after 28 years, this is not a temporary status," and that continuation of the decree will only increase disparity in favor of minorities at the entry level. Claimants argue that this changed circumstance implicates narrow tailoring because the Beecher decree has proven ineffective in achieving its goals. In other words, the unexpected barrier between the entry-level and the upper ranks, experienced by minorities within the BFD, demonstrates that the race-conscious hiring process is not narrowly tailored to serve the decree's goals of parity within the BFD. — 94 — Interestingly, the majority's narrow tailoring argument and the Candidates's narrow tailoring argument are based on contradictory factual premises. The majority rejects the inclusion of higher ranks in the computation of parity because, in part, such inclusion unconstitutionally transforms the decree into an instrument of "general leverage favoring minorities" in promotions. By contrast, the Candidates assert that the decree is unconstitutional because it is ineffective in furthering the promotion of minorities to the upper ranks. — 95 — There are two ways to understand the Candidates' ineffectiveness argument. The Candidates may be arguing that the Beecher decree, as originally conceived, requires parity at both the entry-level and among the upper ranks. Because of the alleged barrier to the promotion of minority firefighters, the Candidates contend that the decree is ineffective in achieving this goal. Alternatively, they may be arguing that the purpose of the decree is to achieve parity in the department as a whole, irrespective of the distribution of minorities at all levels of the department. Given the barrier to promotions, they assert that the only way this parity can be achieved is indirectly — that is, by impermissible "overloading" of minorities at the entry-level to "offset" the lack of diversity in the upper ranks — and that this effect is fatal to narrow tailoring. I will address separately these variations of the ineffectiveness argument. — 96 — 1. Parity at the Entry Level and in the Upper Ranks — 97 — The Candidates's first ineffectiveness argument is premised on their characterization of the original purpose of the decree, described as an assumption. That is, they attribute to the district court in Beecher the assumption that "once the discriminatory effects of the entrance hurdles were removed, minorities would trickle through the fire department somewhat evenly." There is a basic flaw in this premise. At the time the Beecher decree was entered, the percentage of minority firefighters in the BFD was less than one percent. It would have been pointless for the Beecher court to make any assumptions about promotion practices when the total number of minorities in the BFD was negligible. Moreover, the court did not enter its decree only to remedy discrimination in the Boston fire department. The Beecher court made fifty-five other municipalities in Massachusetts subject to the Beecher decree, and specifically noted the disproportionately low percentage of minorities in the "total force" of other towns.17 Essentially, there were few minorities to promote in any of the cities or towns, and the Candidates never explain why the Beecher court would have made any assumptions about promotion practices. Therefore, it is unpersuasive to attribute to the Beecher court the assumption described in the Candidates' brief that "once the discriminatory effects of the entrance hurdles were removed, minorities would trickle through the fire department somewhat evenly." Instead, it is more reasonable to attribute to the Beecher court the goal reflected in the explicit language of the decree: "a complement of minorities commensurate with the percentage of minorities" within each municipality. Beecher, 371 F.Supp. at 523. The City, an original party to the decree, states the purpose of the decree well in its brief: "increasing the total gross proportion of minority fire personnel in each appointing authority." — 98 — Indeed, as noted earlier, the Beecher court discusses the purpose of the decree in its original opinion ordering remedial action and establishing the framework for the subsequent consent decree. There it states that in 1974 "blacks and Spanish-surnamed persons represent such an insignificant percentage of the force," and that these "present effects of past discrimination must be remedied." Beecher I, 371 F.Supp. at 519-20. The decree says nothing about anticipating racial proportionality in both the entry-level and the upper ranks. The Candidates' argument of ineffectiveness in achieving such proportionality fails because it is premised on a goal that is nowhere discernible in the Beecher decree or decision. 2. Parity in the Department as a Whole — 99 — Even if the purpose of the Beecher decree is to diversify the entire department, without regard to the even distribution of minorities in the entry level and upper ranks, the Candidates argue that, given the barrier to minority promotion in the department, continuation of the decree will only increase disparity in favor of minorities at the entry level. From this changed circumstance, the Candidates offer a complicated narrow tailoring argument that does not lend itself to easy summary. Hence, I quote it at length: — 100 — If the method of promotion is appropriate, is not discriminatory, and nevertheless leads to racial disparity at the top levels, then it is inappropriate to try to "offset" legitimate promotions with an increased minority presence at the bottom. In other words, parity may currently exist in the upper levels of the Fire Department, in that it may reflect the appropriate proportion of minorities in the qualified labor pool for such promotions.... If so, this parity can not be used to justify further hiring quotas in the lower ranks. — 101 — In the alternative, if lack of minorities at the top levels is the result of a discriminatory promotional procedure (of which there is no proof), the requirement to "narrowly tailor" racial preferences requires that the specific promotional problem be addressed directly, and not indirectly through affirmative action at the lowest level. Indeed, more minorities in a lower position does not remedy the problem of a "discriminatory" promotional process. — 102 — To comply with strict scrutiny, a racial remedy must be effective. It is inappropriate to address alleged disparity in the upper ranks indirectly using racial preferences to fill the lower ranks, when minorities are not being promoted in proportionate numbers. — 103 — The Candidates's narrow tailoring argument asserts that there are only two possible causes for the discrepancy between the percentage of minorities at the entry-level and the percentage of minorities in the upper ranks: (a) the low percentage of minorities in the upper ranks is an accurate reflection of the qualified applicant pool, or (b) the low percentage of minorities reflects discrimination in the promotion process. They say that if the barrier to promotion is due to the lack of qualified minority applicants, then the decree is not narrowly tailored because it requires the overloading of minorities at the entry-level to offset this dearth of minorities qualified for the upper ranks. Alternately, if the barrier to promotion is due to discrimination, the decree is ineffective in addressing this problem because it focuses only on hiring practices. Since there are only two possible causes of the discrepancy between the percentage of minorities at the entry-level and the percentage of minorities in the upper ranks, and each cause demonstrates that the decree is not narrowly tailored to serve the decree's goals of parity within the entire Fire Department, the Candidates argue that this "conundrum," as they describe it, conclusively proves that the decree fails the narrow tailoring test. — 104 — This logic reveals another flawed premise. Causes (a) and (b) do not represent the universe of possible explanations for the discrepancy. In fact, the City offers a third plausible and constitutionally sound explanation. Based on its experience with public employment affirmative action programs, the City posits that it may just take a long time for affirmative action measures at the entry level of the BFD to influence minority representation in the higher ranks. Instead of being evidence of a barrier to promotion, the fact that only 6.2% of the higher ranks of the department is minority may evidence an inescapably slow process of growing diversity throughout the department. At the time the Beecher decree was implemented in 1974, minorities represented only 0.9% of the total BFD. Beecher I, 371 F.Supp. at 514. While far from representative of the minority population of the community, the 6.2% minority representation in the upper ranks is a significant increase indicating that the decree, however slowly, is fostering diversity in the upper ranks. Indeed, there has been a demonstrable increase in the number of minority officers recently. The City's response to the Candidates' interrogatories, included in the summary judgment record, indicates that there were 17 minority officers in 1998, 18 in 1999, and 22 in 2000. Over a span of two years, the percentage of minority officers increased from 4.8% to 6.2%. — 105 — Moreover, the slow rate of change in the officer ranks is not surprising in an organization the size of the BFD. While the department employs over 1500 firefighting personnel, only 350 of those hold officer positions — less than 25 percent. Although the record reveals that it is not uncommon for Boston to hire 50 new firefighters in a particular year, it is highly unlikely that a department with only 350 officer positions will promote that many over the same time period. Because the annual promotion rate will generally be much lower than the annual hiring rate, evidence that minorities are not penetrating the upper levels of the department as quickly as the entry level is predictable. In concluding that there is an intractable barrier to minority promotion in the department merely because there are only 22 minority officers, the Candidates take too grudging a view of progress. — 106 — With their conundrum argument, the Candidates have constructed an argument designed to succeed on the basis of logic instead of proof: because both of the possible explanations for the discrepancy between the percentages of minorities in the entry-levels and upper ranks indicate a lack of narrow tailoring, they assert that they do not have to prove the reality of either explanation. However, as noted, the logic is faulty because it fails to account for other plausible and constitutionally sound explanations for the discrepancy. Therefore, even on the assumption that there might be a narrow tailoring problem with the Beecher decree if either of the Candidates' explanations of the discrepancy were real (unqualified applicants for promotion or discriminatory promotion practices), the Candidates must offer proof of one of these explanations. They have failed to do so. — 107 — As the Candidates themselves point out, there is no evidence of discriminatory promotion practices. The alternative "unqualified applicants" explanation18 is objectionable on a number of grounds. First, the Beecher court found that the entrance examination for firefighters was not indicative of job performance and had a discriminatory impact on minorities. Beecher 371 F.Supp. at 517-18. Second, as we made clear in Mackin, 969 F.2d at 1278, the Beecher court specifically designed the decree to require the hiring of only qualified applicants. Third, the Candidates themselves assert that the entry-level and promotional positions have "markedly different skill, experience and knowledge requirements," making performance on the entrance examination of little relevance to promotional success. There is simply no proof in this record that the 6.2% statistic represents the "appropriate proportion of minorities in the qualified labor pool for such promotions." — 108 — The Candidates bear the burden of proving the existence of changed circumstances undermining the constitutionality of a consent decree. Rufo, 502 U.S. at 383, 112 S.Ct. 748. Mere recitation of the current statistics describing the racial composition of the entry-level and upper ranks is insufficient. Thus, the Candidates have failed to carry their burden of proving substantial changed circumstances affecting the constitutional validity of the decree. B. Duration of the Decree — 109 — Although we rejected this argument when we decided Mackin, the Candidates argue that the passage of time (now twenty eight years) is itself a changed circumstance that demonstrates the ineffectiveness of the Beecher decree. That argument is as unpersuasive now as it was when we decided Mackin. Then we said: — 110 — [T]he decree's life is limited, remaining in force only until its requirements have been met.... Indeed, the proof of the present pudding is that, since 1974, more than fifty percent of the communities originally affected by the decree have already been freed from further oversight. — 111 — Mackin, 969 F.2d at 1278 (citations omitted). — 112 — Today, the fifty percent figure cited in Mackin ten years ago has reached eighty percent, with forty-five out of fifty-six communities originally affected by the decree having now achieved parity and thus exclusion from further oversight. The Beecher decree remains limited in duration because it is extinguished upon the achievement of a demonstrably attainable rough parity. See Beecher, 371 F.Supp. at 523 (providing for release from appointment process mandated by the decree "as a city or town achieves a complement of minorities commensurate with the percentage of minorities within the community"). Limitations of this sort are crucial factors in evaluating overbreadth challenges. See Stuart v. Roache, 951 F.2d 446, 454 (1st Cir.1991). Nothing about the race-conscious appointment process of the Beecher decree resembles "remedies that are ageless in their reach into the past, and timeless in their ability to affect the future." Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 276, 106 S.Ct. 1842, 90 L.Ed.2d 260 (1986); see also City of Richmond v. J.A. Croson Co., 488 U.S. 469, 498, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989). With so much compliance elsewhere, and with so much progress in Boston towards rough parity, the mere passage of time does not demonstrate the decree's ineffectiveness. IV. Conclusion — 113 — At least since our decision in Mackin ten years ago, the Beecher decree has had a settled meaning. Pursuant to that settled meaning, the fire departments of eighty percent of the municipalities subject to the Beecher decree have achieved compliance with it. Now, faced with a renewed challenge to the decree by Candidates who feel themselves unfairly barred from employment opportunities in the BFD, the majority finds new meaning in an old decree. I disagree with that conclusion. It cannot be reconciled with our precedent in Mackin, sound principles of interpretation applicable to a consent decree, or the history of this case. Taking a different tack, the Candidates challenge the Beecher decree by constructing changed circumstances arguments that are unpersuasive because of faulty logic, the absence of proof, and prior rejection in Mackin. — 114 — The Beecher decree is not just about the BFD. It was and remains a decree applicable to other cities and towns in Massachusetts. Although it is regrettable that the substantial progress of the BFD in ending its discriminatory hiring practices has not yet reached the goal of rough parity required by the decree, that remaining gap is no justification for altering a decree which requires, with a formula now long accepted, that the "present effects of past discrimination must be remedied." Beecher, 371 F.Supp. at 520. Until that gap is closed, the remedial purpose of the decree remains unfulfilled. We should stay the course, not abandon it. I respectfully dissent. — 115 — Before BOUDIN, Chief Judge, STAHL, Senior Circuit Judge, TORRUELLA, SELYA, LYNCH*, LIPEZ, and HOWARD, Circuit Judges. — 116 — The panel of judges that rendered the decision in this case having voted to deny the petitions for rehearing with one judge dissenting and the suggestions for rehearing en banc having been carefully considered by the judges of the Court in regular active service and a majority of said judges not having voted to order that the appeal be reheard by the Court en banc, — 117 — It is ordered that the petitions for rehearing and the suggestions for rehearing enc banc be denied. — 118 — Chief Judge Boundin, Judge Torruella and Judge Lipez would grant the petitions for rehearing en banc. A separate statement by Judge Lipez, joined by Judge Torruella, is attached to this order. — 119 — LIPEZ, Circuit Judge, with whom TORRUELLA, Circuit Judge, joins, dissenting from the denial of the petitions for rehearing en banc. — 120 — I have already explained in a dissent my disagreement with the panel majority. I will not repeat those reasons here. However, without suggesting any disrespect for the majority's views, I wish to record my vote for rehearing en banc and state my concern about any possible implications of the majority's reliance on the narrow tailoring doctrine. That reliance does not seem central to the majority's holding. Moreover, given the close vote on the en banc petitions, that reliance may have limited value for any future litigation involving the application of the narrow tailoring doctrine to a race-conscious remedial decree. Footnotes [1] Notes: [1] Throughout the litigation, the parties have used the term "blacks" as opposed to "African-Americans." For historical coherence, we henceforth will adhere to that usage. For ease in reference, however, we elect to use the term "Hispanics" as a proxy for the more cumbersome "Spanish-surnamed individuals." Finally, we use the words "minority" and "minorities" to refer to blacks and Hispanics collectively, and the word "non-minority" to encompass all other persons [2] The parties skirmish over both the construction of the eligibility list and what deviations, if any, the BFD entertained from it. Although these disputes ultimately may prove material in addressing the relief to which the Candidates may be entitled,see infra Part III(E), we leave them unresolved for purposes of this appeal. [3] These counts are not in issue on this appeal,see infra Part II, and we therefore omit any detailed discussion of them. [4] By way of illustration, the BFD had a minority complement at the time of less than 1%, while Boston had a minority population of approximately 23%Beecher I, 371 F. Supp. at 514. [5] Our dissenting colleague chastises us for giving the Candidates too much credit. He says that the Candidates do not argue that the parties' original intent encompassed only the entry-level rank. That is literally true, but it overlooks the fact that the Candidates place the scope of the decree squarely at issue and argue that constitutional principles require such an interpretation. To address that argument, our analysis necessarily begins with a "determination of the parties' intent when they entered into the stipulation."Navarro-Ayala, 951 F.2d at 1339. [6] Throughout, all references to "entry-level" positions should be understood to mean post-probationary non-officer positions. This qualification is compelled by explicit language in theBeecher decree enjoining "further certifications of permanent appointments ... to the position of firefighter" until certain requirements of the decree are met. Beecher I, 371 F.Supp. at 521 (emphasis supplied). That reference is reinforced by language in the decree that differentiates between permanent and provisional appointments. Id. [7] Beecher was a consolidated set of class actions in which the district court certified two classes. The first class consisted of "[a]ll black or Spanish-surnamed persons who have applied for the position of firefighter in any fire department ... subject to Massachusetts Civil Service law, but have not become eligible for appointment under existing requirements." Beecher I, 371 F.Supp. at 510. The other class included "[a]ll black or Spanish-surnamed persons who have never applied for the position of firefighter because they have allegedly been deprived of information concerning firefighter employment opportunities as a result of the allegedly discriminatory recruitment practices of the defendants." Id. [8] This statute currently imposes an upper age limit (32) for new firefightersSee Mass. Gen. Laws ch. 31, § 58. Because no maximum age applied to the November 2000 hiring cycle, we do not dwell on this feature. [9] We note that the district judge who actually entered theBeecher decree (Judge Freedman) made specific mention that, in 1974, the BFD required applicants to be within an age range of 19-35 years. Beecher I, 371 F.Supp. at 512. He also observed that, prior to 1971, the minimum age had been 21. Id. Mindful of this shift, he concluded that the community population as a whole would be an appropriate measure of compliance. This seems supportable in light of Judge Freedman's recognition that the enjoined party had the power to reconfigure the contours of the labor pool. The authorities cited by the Candidates involve parties without such leverage vis-à-vis the reach of the remedy; where such leverage exists, however, measuring parity against general population statistics prevents a public employer enjoined for discriminatory practices in violation of federal law from manipulating the remedy to suit its fancy. So viewed, Judge Freedman's shaping of the second variable is a good example of why "broad judicial discretion [is often] crucial for the district judge to secure complex legal goals." Navarro-Ayala, 951 F.2d at 1338 (citations and internal quotation marks omitted). [10] Our dissenting brother never squarely addresses the Candidates' argument that "even assuming that general population statistics are to be considered, the minority composition of the Department's firefighters (39.9%) exceeded the minority composition of the general population of Boston (38.3%). The examination of parity should focus on minority representation in the firefighter position alone .... The racial preferences at issue apply solely to the firefighter position." Appellants' Brief at 23 (internal cross-references and paragraph structure omitted). This is a changed circumstance requiring an examination into "whether the goals of the litigation ... have been completely achieved,"Mackin, 969 F.2d at 1275, as well as a question posed for the first time involving the decree's constitutional application. Viewed in either fashion, the question warrants strict scrutiny. See Rufo v. Inmates of Suffolk County Jail, 502 U.S. 367, 384, 112 S.Ct. 748, 116 L.Ed.2d 867 (1992); see also Adarand, 515 U.S. at 235, 115 S.Ct. 2097. Our dissenting brother, however, says only that he "cannot imagine that the plaintiffs in the Beecher litigation would have decided upon a remedy that provides incentives [to keep minorities in the entry-level rank]." This is the stuff not of strict scrutiny but of rational basis review. See, e.g., FCC v. Beach Comm'ns, Inc., 508 U.S. 307, 315, 113 S.Ct. 2096, 124 L.Ed.2d 211 (1993) (upholding law under rational basis review "based on rational speculation unsupported by evidence or empirical data"). [12] Notes: [11] In our first review of the decree, we paraphrased the parity paragraph of the decree, saying that "[t]he decree remains in force, for each local fire department, until that department attains sufficient minority fire fighters to have a percentage on the force approximately equal to the percentage of minorities in the locality."Boston Chapter, NAACP, Inc. v. Beecher, 504 F.2d 1017, 1026-27 (1st Cir. 1974). While the use of the term "fire fighters" may have been a shorthand description of the decree, it certainly should not supplant the language of the original decree. [12] The majority dismisses the city's reliance on the terminology of the November 1975 Agreement because that Agreement "incorporates by reference paragraph 7" of the April 1975 decree, which uses the term "firefighters," and therefore "leads us back to our starting point." The majority assumes that the April 1975 decree should be the court's "starting point." The starting point of the analysis should be the originalBeecher decree — which uses the phrase "complement of minorities." Beecher, 371 F.Supp. at 523. [13] In support of its argument that "firefighter" is the operative term for interpreting the decree, the majority notes that the term "firefighter" is used "no fewer than twenty-six times in the decrees, while the term `uniformed personnel' appears only once." While likely true, this statement is beside the point. For example, the term "firefighter" is used no fewer than ten times simply to describe the examination given to candidates, for example "valid firefighter entrance examination" and "fire fighter examination." As I indicated, one must focus on the operative term of the decrees, i.e. the language which explains the meaning of parity. In that context, the word "firefighter" is one of only three formulations [14] The bracketed words are in the April 1975 decree [15] Although theBeecher court referred to the Division of Personnel Administration, the agency calls itself the Department of Personnel Administration. I assume that the agencies in question are one and the same. There does not currently exist a Massachusetts agency with the official name Division of Personnel Administration. [16] In a footnote to its opinion, the majority cites this point as evidence that I do not understand the difference between strict scrutiny and rational basis review. Self-evidently, I make this point only in support of the original meaning of the decree advanced by the defendants and acknowledged by the Candidates. I do not proffer it as a constitutional basis for upholding the decree. My conclusion that this original interpretation of the decree is constitutional, or more specifically, that it is narrowly tailored to a compelling government interest, is presented in the section that follows [17] "The City of Springfield has a black population of approximately 13%. Of 475 firefighters in that city, one is black and none are Spanish-surnamed. The one black represents 0.2% of thetotal force. The City of Cambridge has a black population of 6.1%, and a fire fighting force of 305 men, four of whom are black; one is Spanish-surnamed. They represent approximately 2% of the total force. New Bedford has a black population of approximately 3.5%. Minorities represent approximately 1% of the fire force. Worcester has a black population of approximately 2%. Minorities represent 1% of the fire force there." Beecher, 371 F.Supp. at 514 (emphasis added). [18] The Candidates elaborate on this explanation as follows: "[B]ecause affirmative action benefits minorities that score lower on the entrance examinations ... [w]hen these lower scorers seek promotions, they are competing against non-minorities who have scored at the very top of the entrance examination." [-s] * Judge Lynch is recused and did not participate in the consideration of this matter