The Little Rock School District (LRSD) appeals from the district court's1 denial of its request for unitary status. We affirm.

LRSD has been involved in federal desegregation litigation since 1956; the present phase of this case commenced in 1982. See Little Rock Sch. Dist. v. Pulaski County Special Sch. Dist. No. 1, 778 F.2d 404 (8th Cir.1985) (en banc); see also Little Rock Sch. Dist. v. Pulaski County Special Sch. Dist. No. 1, 921 F.2d 1371, 1376-83 (8th Cir.1990) (chronicling litigation history); Polly J. Price, The Little Rock School Desegregation Cases in Richard Arnold's Court, 58 Ark. L.Rev. 611, 622-47 (2005). In 1989, the district court approved an interdistrict settlement plan (1989 Settlement), which specified that the district court would supervise the remedial desegregation efforts of LRSD and two neighboring school districts. We ordered the creation of the Office of Desegregation Monitoring (ODM) to assist the district court in its supervision. See Little Rock Sch. Dist. v. Pulaski County, 921 F.2d at 1388, 1394.

By 1996, it had become apparent to the parties and the district court that LRSD would be unable to meet the terms of the 1989 Settlement. At the suggestion of the district court, LRSD and Lorene Joshua (Joshua), the class representative for all African-American students enrolled in LRSD and the two neighboring districts, entered into negotiations to modify LRSD's obligations. The fruit of these negotiations was the Revised Desegregation and Education Plan (Revised Plan), which the district court approved in 1998. It was agreed that if LRSD substantially complied with the terms of the Revised Plan, it would be declared unitary at the conclusion of the 2000-2001 school year. The Revised Plan required Joshua to submit any unresolved compliance issues to the ODM for facilitation of an agreement. Revised Plan § 8.2.4. If the ODM could not resolve the issue "after good faith attempts at facilitation," Joshua could seek resolution of the issue with the district court. Id. § 8.2.5.

In 2002, the district court granted LRSD partial unitary status, finding that it had complied with all but section 2.7.1 of the Revised Plan. Little Rock Sch. Dist. v. Pulaski County Special Sch. Dist., 237 F. Supp. 2d 988, 1089 (E.D.Ark.2002). Section 2.7.1 provided that:

LRSD shall assess the academic programs implemented pursuant to Section 2.7 after each year in order to determine the effectiveness of the academic programs in improving African-American achievement. If this assessment reveals that a program has not and likely will not improve African-American achievement, LRSD shall take appropriate action in the form of either modifying how the program is implemented or replacing the program.

Compliance Plan, Oct. 10, 2002, at 1 (footnote omitted). The district court imposed a compliance remedy (2002 Remedy) designed to bring LRSD into substantial compliance with section 2.7.1. Id. at 1087-88. On appeal by Joshua from the district court's unitary-status ruling, we affirmed. Little Rock Sch. Dist. v. Armstrong, 359 F.3d 957 (8th Cir.2004).

On March 15, 2004, following what it believed was its substantial compliance with section 2.7.1 and the 2002 Remedy, LRSD asked to be declared unitary. Joshua opposed the request. On June 30, 2004, the district court concluded that LRSD had not substantially complied with its obligations, denied unitary status, and imposed a new compliance remedy (2004 Remedy). It is from this judgment that LRSD now appeals.

Although this case traces its roots to federal desegregation efforts, the instant appeal presents no constitutional issues. The constitutional requirements for unitary status are set forth in Green v. County School Board, 391 U.S. 430, 435-38, 88 S. Ct. 1689, 20 L. Ed. 2d 716 (1968), which held that a school district may be declared unitary and lacking racial discrimination based on satisfactory performance in five areas of a school district's operations: (1) student assignment; (2) faculty and staff assignment; (3) transportation; (4) extracurricular activities; and (5) facilities. LRSD has met these requirements. See 237 F.Supp.2d at 1089 (declaring LRSD "partially unitary with regard to all aspects of its operations, because it has substantially complied with all sections of the Revised Plan, save for those obligations contained in § 2.7.1"). As the district court noted in 2002, the Revised Plan "included other desegregation obligations that went beyond Green's constitutional floor for a school district to become unitary." Id. at 1032.

Section 2.7.1 of the Revised Plan refers to section 2.7, which required LRSD to "implement programs, policies and/or procedures designed to improve and remediate the academic achievement of African-American students, including but not limited to Section 5 of this Revised Plan." The impetus for section 2.7 is colloquially referred to as the "achievement gap" between minority students and white students in the public school systems. For section 2.7.1 to meet the constitutional threshold of a desegregation remedy, the achievement gap that it was designed to remedy would have to "directly address and relate to the constitutional violation itself." Missouri v. Jenkins, 515 U.S. 70, 88, 115 S. Ct. 2038, 132 L. Ed. 2d 63 (1995); see also Bd. of Educ. v. Dowell, 498 U.S. 237, 248, 111 S. Ct. 630, 112 L. Ed. 2d 715 (1991) ("The legal justification for displacement of local authority by an injunctive decree in a school desegregation case is a violation of the Constitution by the local authorities."). As the district court observed in 2002:

[N]o court has ever determined generally, or with the specificity required in [Jenkins], what portion, if any, of the minority student achievement gap in LRSD is causally linked as a vestige of de jure segregation. Furthermore, Joshua has failed to introduce any evidence to establish that: (1) the achievement gap is causally linked to the system of de jure segregation which existed in LRSD decades earlier; and (2) the portion of the achievement gap which is attributable to de jure segregation, after excluding all of the socioeconomic factors that also have contributed to that gap.

237 F.Supp.2d at 1040. Cf. People Who Care v. Rockford Bd. of Educ., 246 F.3d 1073, 1076 (7th Cir.2001) ("[I]t is obvious that other factors besides discrimination contribute to unequal educational attainment, such as poverty, parents' education and employment, family size, parental attitudes and behavior, prenatal, neonatal, and child health care, peer-group pressures, and ethnic culture."). Accordingly, it is unclear whether LRSD's efforts to remedy the achievement gap are constitutionally compelled. We need not determine that issue, however, because LRSD's obligations under section 2.7.1 are clearly contractual matters. Thus, we examine LRSD's compliance under ordinary rules of contract interpretation.

LRSD's obligations under section 2.7.1 therefore arise as a matter of contract, not constitutional law, and thus we examine LRSD's compliance under ordinary rules of contract interpretation. United States v. Knote, 29 F.3d 1297, 1299 (8th Cir.1994). We review the district court's factual findings for clear error, Little Rock Sch. Dist. v. Pulaski County Special Sch. Dist., 83 F.3d 1013, 1017 (8th Cir.1996), and its modification of a consent decree for an abuse of discretion. McDonald v. Armontrout, 908 F.2d 388, 390 (8th Cir.1990). We review de novo the district court's interpretation of the terms of the Revised Plan. Armstrong, 359 F.3d at 965.

This litigation has been complicated by the shifting terminology employed by LRSD, Joshua, and the district court. Specifically, efforts to resolve this case since the Revised Plan took effect in 1998 have been marked by confusion over four terms: (1) assessment; (2) evaluation; (3) program; and (4) key program. The plain language of section 2.7.1 of the Revised Plan, quoted in Part I, supra, required LRSD to assess academic programs implemented pursuant to section 2.7, including but not limited to programs implemented under section 5.

On March 15, 2000, LRSD submitted an Interim Compliance Report that described how it would meet its obligations under the Revised Plan. The report identified myriad programs that had been implemented pursuant to section 2.7 and section 5,2 and set forth both a "Program Evaluation Agenda"3 and an "Assessment Plan" in response to section 2.7.1. Interim Compliance Report at 53-57. Significantly, the Program Evaluation Agenda indicated that LRSD would evaluate (rather than assess) only a small subset of the section 2.7 and section 5 programs identified in the Interim Compliance Report, and the Assessment Plan referred only to assessing students (i.e., testing), not programs. Id. LRSD submitted its final Compliance Report on March 15, 2001. The Compliance Report incorporated by reference the Interim Compliance Report. It contained additional information responsive to section 2.7.1 under the title "Program Evaluation," which indicated that LRSD had evaluated fourteen programs.4 Compliance Report at 148.

As recounted above, the district court granted LRSD partial unitary status in 2002, but concluded that LRSD had not substantially complied with the requirements of section 2.7.1. The district court noted that the Interim Compliance Report had "identified almost 100 programs that [LRSD] had implemented to `improve and remediate the academic achievement of African-American students'" under sections 2.7 and 5 of the Revised Plan. 237 F. Supp. 2d at 1076 n. 135. According to the district court, these were "all of the programs... that LRSD was implementing in accordance with its obligations under the Revised Plan." Id. at 1018. The district court elaborated that:

Section 2.7.1 of the Revised Plan required LRSD to assess annually each of the academic programs promulgated pursuant to § 2.7, in order to determine if those programs, in fact, were improving African-American achievement. If the assessment of a program revealed that it was not effective in improving African-American achievement, LRSD was required to modify or replace the program. I find that the purpose of § 2.7.1 was to make sure that the programs promised under § 2.7 actually worked to improve the academic achievement of African-American students. I further find that LRSD's substantial compliance with § 2.7.1 was crucial to its commitment to improve the academic achievement of African-American students; for, without performing a rigorous annual assessment of each of the many dozens of programs implemented under § 2.7, it would be impossible to determine which programs were working and should be continued and which programs were not working and should be discontinued, modified, or replaced with new programs.

Id. at 1076 (footnote omitted).

The above-cited passage is consistent with the terms of the Revised Plan. Elsewhere in its order, however, the district court observed that LRSD's Interim Compliance Plan had specified that evaluations would be conducted for key programs including (1) Pre-K-3 Literacy Program Success For All; (2) the implementation of the new curriculum in English language arts, mathematics, and science in the middle schools; (3) the National Science Foundation project aimed at improving African-American achievement in math and science; and (4) the implementation of the School Improvement Plan. Id. at 1077. The district court found "that the much more in-depth and analytical program evaluations, which LRSD agreed to prepare on certain key remediation programs identified in the Interim Compliance Report, were an integral and essential part of LRSD's obligation under § 2.7.1." Id. at 1079.

The district court also noted the distinction between "assessment" and "evaluation." According to testimony from Dr. Bonnie Lesley, LRSD's then-Associate Superintendent of Curriculum and Instruction, an assessment is "dynamic, it is interactive, it's ongoing, it happens frequently, and it is a measurement, along with the analysis that you would make of whatever results are available." Id. at 1077. In contrast, an evaluation is "more long term, [it] may consider observations or measurements in addition to test scores, and is guided by a set of research questions that are usually provided by whoever the consumer is of that report." Id. The district court noted that LRSD had interpreted section 2.7.1 "to include an obligation to perform some program evaluations." Id.

At the conclusion of its order, the district court set forth the 2002 Remedy, which contained six principal sections. The first four sections are those which are the most relevant to the issue before us:

A. For the entire 2002-03 school year and the first semester of the 2003-04 school year, through December 31, 2003, LRSD must continue to assess each of the programs implemented under § 2.7 to improve the academic achievement of African-American students. LRSD now has over three years of testing data and other information available to use in gauging the effectiveness of those programs. I expect LRSD to use all of that available data and information in assessing the effectiveness of those programs and in deciding whether any of those programs should be modified or eliminated.

B. LRSD must maintain written records regarding its assessment of each of those programs. These written records must reflect the following information: (a) the written criteria used to assess each program during the 2002-03 school year and the first semester of the 2003-04 school year; (b) the results of the annual assessments of each program, including whether the assessments resulted in program modifications or the elimination of any programs; and (c) the names of the administrators who were involved with the assessment of each program, as well as at least a grade level description of any teachers who were involved in the assessment process (e.g., all fourth grade math teachers; all eighth grade English teachers, etc.).

C. LRSD must use Dr. Nunnerly [sic] or another expert from outside LRSD with equivalent qualifications and expertise to prepare program evaluations on [fourteen specified programs].... [A]s these program evaluations are prepared, LRSD shall use them, as part of the program assessment process, to determine the effectiveness of those programs in improving African-American achievement and whether, based on the evaluations, any changes or modifications should be made in those programs. In addition, LRSD must use those program evaluations, to the extent they may be relevant, in assessing the effectiveness of other related programs.

D. Joshua must monitor LRSD's compliance with § 2.7.1 and must immediately bring to the attention of LRSD all problems that are detected in its compliance with its obligations under § 2.7.1, as those obligations are spelled out in this Compliance Remedy. Thereafter, Joshua and LRSD must use the "Process for Raising Compliance Issues" set forth in § 8.2, et seq., of the Revised Plan to attempt to resolve those compliance issues. If those efforts are unsuccessful, Joshua must present the issues to me for resolution, as required by § 8.2.5. Any such presentation must be timely.

Id. at 1087-88 (footnote omitted).

The contrast between the assessments of Subparts A and B and the evaluations of Subpart C demonstrates that the district court was making a clear distinction between the two terms. Viewed in this light, the meaning of Subparts A and B is that the district court was requiring assessments for "each of the many dozens of programs implemented under § 2.7," id. at 1076, irrespective of the requirements imposed by Subpart C. This interpretation is consistent with the district court's order as a whole, the Revised Plan, and LRSD's Interim Compliance Report. Cf. Little Earth of United Tribes, Inc. v. United States Dep't of Hous. & Urban Dev., 807 F.2d 1433, 1438, 1439 (8th Cir.1986) (We view an earlier order by the district court "as a whole, and an interpretation that gives effect to all parts of the order will be preferred over one that leaves portions of the order meaningless or insignificant.... If there is any ambiguity in the ... order, its meaning is clarified by `what preceded it and what it was intended to execute.'").

On October 10, 2002, LRSD's Board of Directors adopted a Compliance Plan that outlined how LRSD would satisfy the 2002 Remedy. The Compliance Plan indicated that LRSD would "[c]ontinue to administer student assessments through the first semester of 2003-04" and "[m]aintain written records of ... the results of the annual student assessments, including whether an informal program evaluation resulted in program modifications or the elimination of any programs." Compliance Plan at 3. These statements reflected LRSD's interpretation of the Revised Plan by referring to LRSD's efforts to administer "student assessments" rather than to its obligation to "assess academic programs."

The Compliance Plan also stated that LRSD would "[p]repare a comprehensive program evaluation of each academic program implemented pursuant to Revised Plan § 2.7 to determine its effectiveness in improving the academic achievement of African-American students and to decide whether to modify or replace the program," and indicated that LRSD would satisfy this obligation by preparing "the following new, comprehensive evaluations: (a) Elementary Literacy, (b) Middle and High School Literacy and (c) K-12 Mathematics and Science." Id. at 3, 5. LRSD thus substituted "comprehensive program evaluations" for the Revised Plan requirement of "assessments" and, more significantly, construed "each academic program implemented pursuant to Revised Plan § 2.7" to mean three broad-based programs. LRSD provided a copy of its Compliance Plan to Joshua in October 2002. See October 25, 2002, letter from John C. Fendley, Jr. (App.2168-72). Joshua disagreed with the proposal outlined in the plan and, pursuant to the Revised Plan, requested that the ODM facilitate resolution of the disagreement. Although the facilitation apparently failed, Joshua did not contact the district court regarding the disagreement.

On March 15, 2004, LRSD submitted its new Compliance Report (2004 Compliance Report). LRSD indicated that its obligation to "[c]ontinue to administer student assessments" had been satisfied through its implementation of "the 2002-03 Board-approved assessment plan." 2004 Compliance Report at 2-3. With respect to how it had met its self-described requirement to "prepare a comprehensive program evaluation of each academic program implemented pursuant to Revised Plan § 2.7," LRSD offered the following explanation:

The LRSD contracted with Dr. Steve Ross, an expert approved by Joshua, to prepare comprehensive evaluations of the District's elementary and secondary literacy programs. These evaluations, combined in a single report, were completed and approved by the Board in November of 2003.... Dr. Don Wold, a program evaluator funded through a National Science Foundation ("NSF") grant; Dennis Glasgow, Interim Associate Superintendent for Curriculum and Instruction; and Vanessa Cleaver, Director of the NSF Grant, authored the comprehensive mathematics and science evaluation. The comprehensive mathematics and science evaluation was completed and approved by the Board in December 2003.

Id. at 3-4.

On June 30, 2004, the district court concluded that LRSD had not substantially complied with the Revised Plan or the 2002 Compliance Remedy. The district court commented that:

While the fields of "Literacy" and "Math and Science" may be convenient ways to divide academic knowledge, they most certainly do not constitute specific § 2.7 "academic programs" (e.g., Reading for All, Early Literacy Learning, Reading Recovery, or Effective Literacy) that LRSD implemented, on a school-by-school basis, to improve the academic achievement of African-American students.

D. Ct. Order of June 30, 2004, at 55. Addressing what it perceived to be LRSD's failure to distinguish between the concepts of assessment and evaluation, the district court explained that:

Subparts A and B of the Compliance Remedy obligated LRSD to assess each of the § 2.7 programs.... I made no mention of LRSD preparing evaluations of § 2.7 programs because, on its face, nothing in § 2.7.1 of the Revised Plan obligated LRSD to perform "program evaluations." However, Dr. Lesley made it clear in her testimony that LRSD administrators knew and understood that the "assessment" obligation in § 2.7.1 included the obligation of preparing "program evaluations." [citation omitted.] Therefore, I concluded it would be best to use the same terms in the [2002] Compliance Remedy that the parties themselves had chosen to use in § 2.7.1 of the Revised Plan.

Id. at 8.

LRSD's 2000 Interim Compliance Report indicated that LRSD construed the Revised Plan's requirement to "assess all programs" to mean that it must "evaluate key programs." Because Joshua elected not to challenge LRSD's interpretation, that interpretation became controlling under ordinary principles of contract law. In its 2002 order, the district court essentially concluded that LRSD had failed to comply substantially with both interpretations of section 2.7.1, i.e., that LRSD had neither assessed all programs nor adequately evaluated key programs. The district court thus imposed a bifurcated compliance remedy. Subparts A and B reverted to the original, plain meaning, requirements of section 2.7.1. Subpart C addressed LRSD's interpretation. Had LRSD appealed, Subparts A and B might well have been deemed to be superfluous requirements. LRSD did not appeal the 2002 order, however, and thus the district court's remedy became the governing interpretation of the section 2.7.1 obligations.

LRSD's 2002 Compliance Plan construed Subparts A and B in the same manner that it had previously interpreted section 2.7.1: it took "assess all programs" to mean "evaluate key programs." When Joshua again failed to raise a legally sufficient challenge to this interpretation, LRSD arguably became entitled to rely on its interpretation, i.e., that three broad-based program evaluations would satisfy the requirements of Subparts A and B.

The district court, however, refused to find that Joshua had waived its right to challenge LRSD's compliance, saying that "[i]n a school desegregation case that has its origins in the infamous 1957 Little Rock school desegregation crisis, no court is likely to hold the silence of Joshua's counsel — even if they are to be criticized — against the African-American students they represent, and who now fill almost 70% of the total number of seats in LRSD's classrooms." D. Ct. Order of June 30, 2004, at 20.

The district court found that LRSD had failed to substantially comply with both the plain meaning of Subparts A and B (i.e., assess all programs) and LRSD's interpretation of the remedy (i.e., evaluate three key programs).

At the outset, we have substantial concerns about the highly detailed, complex nature of the district court's 2002 Remedy, which imposes upon LRSD additional requirements, some of which appear to go well beyond those agreed upon by the parties in the Revised Plan. The Revised Plan constituted a contract between the parties, and the district court was not free to expand its terms beyond that which was contemplated by the parties. Subpart C of the 2002 Remedy exceeded the scope of the Revised Plan, which lacked any requirement for program evaluations. See Krupnick v. Ray, 61 F.3d 662, 664 (8th Cir.1995) ("The law of Arkansas provides that it is the duty of the court to construe the contract according to its unambiguous language without enlarging or extending its terms."); cf. Holland v. N.J. Dep't of Corr., 246 F.3d 267, 281 (3d Cir.2001) ("A court should interpret a consent decree as written and should not impose terms when the parties did not agree to those terms."); Equal Employment Opportunity Comm'n v. N.Y. Times Co., 196 F.3d 72, 78 (2d Cir.1999) ("A court may not replace the terms of a consent decree with its own, no matter how much of an improvement it would make in effectuating the decree's goals."); Armstrong v. Bd. of Sch. Dirs. of Milwaukee, 616 F.2d 305, 315 (7th Cir. 1980) ("Judges should not substitute their own judgment as to optimal settlement terms for the judgment of the litigants and their counsel."). LRSD, however, elected not to appeal Subpart C in 2002, and so that issue is not before us today.

In its June 30, 2004, order, the district court set forth an even more highly detailed compliance remedy (the 2004 Remedy). We cite selected portions of that Remedy as representative of the level of specificity that the district court engaged in:

Because LRSD failed to substantially comply with the crucially important obligations contained in § 2.7.1 of the Revised Plan, it must remain under court supervision for two more complete school years, 2004-05 and 2005-06. To avoid any "misunderstanding" regarding this Compliance Remedy, I will be specific. The new Compliance Remedy is as follows:

A. LRSD must promptly hire a highly trained team of professionals to reinvigorate PRE. These individuals must have experience in: (a) preparing and overseeing the preparation of formal program evaluations; and (b) formulating a comprehensive program assessment process that can be used to determine the effectiveness of specific academic programs designed to improve the achievement of African-American students. I expect the director of PRE to have a Ph.D.; to have extensive experience in designing, preparing and overseeing the preparation of program evaluations; and to have a good understanding of statistics and regression analysis. I also expect LRSD to hire experienced statisticians and the other appropriate support personnel necessary to operate a first-rate PRE Department.

B. The first task PRE must perform is to devise a comprehensive program assessment process. It may take a decade or more for LRSD to make sufficient progress in improving the academic achievement of African-American students to justify discontinuing the need for specific § 2.7 programs. For that reason, the comprehensive program assessment process must be deeply embedded as a permanent part of LRSD's curriculum and instruction program. Only then will I have the necessary assurance that LRSD intends to continue using that process for as long as it is needed to determine the effectiveness of the various key § 2.7 programs in improving the academic achievement of African-American students....

C. During each of the next two academic school years (2004-05 and 2005-06), LRSD must hire one or more outside consultants to prepare four (4) formal step 2 evaluations. Each of these step 2 evaluations must cover one of the key § 2.7 programs, as it has been implemented in schools throughout the district. Thus, over the course of the next two academic years, LRSD must hire outside consultants to prepare a total of eight (8) formal step 2 evaluations of key § 2.7 programs. During the recent compliance hearing, Dr. Ross made it clear that LRSD must conduct these formal step 2 evaluations of the key § 2.7 programs in order to continue to make progress in improving the academic achievement of African-American students. Again, I suggest that LRSD hire Dr. Ross—to perform the following tasks: (1) identify the four key § 2.7 programs that should be formally evaluated during the 2004-05 school year and the four key § 2.7 programs that should be formally evaluated during the 2005-06 school year; and (2) prepare as many of the eight step 2 evaluations as possible. If Dr. Ross cannot prepare all eight of the step 2 evaluations, I recommend that LRSD hire someone that Dr. Ross recommends as possessing the experience and ability necessary to prepare those evaluations.

....

E. In order to streamline LRSD's record-keeping obligation, I am going to require that each of the eight step 2 evaluations contain, in addition to the traditional information and data, a special section which: (1) describes of the number of teachers and administrators, at the various grade levels, who were interviewed or from whom information was received regarding the effectiveness of the key § 2.7 program being evaluated; (2) lists each of the recommended program modifications, if any, that were deemed necessary in order to increase the effectiveness of each of the § 2.7 programs in improving the academic achievement of African-American students; and (3) briefly explains how each of the recommended modifications is expected to increase the effectiveness of the § 2.7 program. This requirement is intended to relieve LRSD of any independent record-keeping obligations under § 2.7.1 of the Revised Plan and the Compliance Remedy.

....

G. PRE must submit quarterly written updates on the status of the work being performed on the four step 2 program evaluations that will be prepared during the 2004-05 school year and the four step 2 program evaluations that will be prepared during the 2005-06 school year. These quarterly updates must be delivered to the ODM and Joshua on December 1, March 1, June 1, and September 1 of each of those two academic school years. As soon as each of the eight step 2 evaluations has been completed and approved by the Board, LRSD must provide a copy to the ODM and Joshua.

....

K. On or before October 15, 2006, LRSD must file a Compliance Report documenting its compliance with its obligations under § 2.7.1 of the Revised Plan, as specified in this Compliance Remedy. If Joshua wishes to challenge LRSD's substantial compliance, they must file objections on or before November 15, 2006. Thereafter, I will schedule a compliance hearing and decide whether LRSD has met its obligations under the Compliance Remedy and should be released from all further supervision and monitoring.

L. This Compliance Remedy is intended to supersede and replace the identical compliance obligations that I imposed on LRSD, albeit with less specificity, in subparts A and B of Section VII of the September 13 Decision.

D. Ct. Order at 61-67 (footnotes omitted). There was much more in the way of detailed requirements, but those portions quoted are illustrative of what it is that LRSD is now being asked to do.

In reviewing the district court's determination that LRSD had not substantially complied with the terms of the Revised Plan and the 2002 Remedy, we note, as did the district court, that on March 14, 2003, the same day that LRSD filed what are termed its "Page 148 Evaluations," Dr. Lesley, the person responsible for overseeing these evaluations, resigned. Two months later, Dr. T. Kenneth James, LRSD's superintendent, also resigned. Thus the two persons to whom direct responsibility had been assigned to ensure compliance with the requirements of the 2002 Remedy were no longer available to assist LRSD in the implementation of the Compliance Plan. Indeed, the authors of the ODM's March 30, 2004, Compliance Report observed that the loss of Dr. Lesley and Dr. James at a crucial time in the implementation of the Compliance Plan, coupled with the delays and difficulties LRSD encountered in filling those positions with acting or interim employees, created "a period of some uncertainty" for LRSD.

We note these facts and observations to highlight the constraints under which LRSD was laboring as it sought both to satisfy the requirements of the Compliance Plan and to demonstrate through empirically based evidence that it had in fact accomplished that goal.

LRSD contends that by failing to challenge the adequacy of LRSD's Compliance Plan, Joshua in effect abandoned its objection to that plan and thus should not be heard to contest LRSD's compliance with the 2002 Remedy. If this case involved only a dispute between private litigants, we might well have disagreed with the district court's holding that Joshua had not waived its right to challenge either LRSD's interpretation of the 2002 Remedy or LRSD's claim that it had substantially complied with the requirements of that remedy. Given the lengthy nature of the litigation between the parties, however, and the substantial likelihood that LRSD's ongoing attempts to comply with the 2002 order will in due course entitle it to be declared unitary, we will not reverse the district court's ruling on this issue. Nevertheless, in light of its failure to call to the district court's attention to its disagreement with LRSD's interpretation of the 2002 order, it would ill behoove Joshua to raise any further technical complaints about LRSD's efforts to comply with the 2002 order.

Without recounting in depth the voluminous evidence that LRSD submitted in support of its claim that it had substantially complied with the Revised Plan and the 2002 Remedy, we observe that had the question of compliance been submitted to us in the first instance, we might well have found that LRSD had met its burden of proof, all the more so in light of the heightened requirements imposed by the district court in its 2002 order. It is a close question whether all of those additional requirements are within the scope of the Revised Plan or whether they represent newly created, after-the-fact fine-tuning that neither of the parties contemplated when they entered into the settlement agreement that resulted in the Revised Plan.

Nevertheless, and once again adverting to the lengthy, if not indeed tortuous, path on which this litigation has proceeded, we conclude that the district court, although it may have come close to crossing the line between proper judicial enforcement of an agreed-upon undertaking and the imposition of requirements that find no warrant in that undertaking, did not clearly err in finding that LRSD had failed to demonstrate substantial compliance with the Revised Plan and the 2002 Remedy.

Our concerns regarding the 2002 Remedy are even greater with respect to the even more heightened requirements, as illustrated by the portions quoted above, imposed by the district court's June 30, 2004, order. One can understand the frustration expressed by LRSD concerning that which it is now expected to do. Indeed, during oral argument we asked LRSD's counsel, "Can you tell us in plain, simple language what it is that [the district court] wants the District to do that it is unwilling to do?" Counsel responded in part by saying that LRSD was complying with the new compliance remedy. In light of that representation, we are unwilling at this time to say that those heightened requirements surpass beyond all measure the requirements to which LRSD committed itself when it entered into the 1989 Settlement. Suffice it to say that there will be time enough for us to revisit the requirements of the 2004 order if this case should once again come before us. For the moment, then, we offer no comment beyond the observation that substantial compliance means just that, not mathematical precision.

In the concluding paragraphs of its June 30, 2004, order under the heading "Final Thoughts on LRSD's Compliance Efforts," the district court, referring to its review of the history of the negotiations that had led the parties to voluntarily enter into the Revised Plan, stated:

I review this history to make it crystal clear that LRSD — not the Court — formulated all of the program assessment/evaluation obligations contained in § 2.7.1 of the Revised Plan and voluntarily agreed to comply with all of those obligations.

I know it will be quite a burden for LRSD to formulate, implement, and deeply embed in its curriculum an effective § 2.7.1 program assessment/evaluation process that will allow it to determine the effectiveness of each of the key § 2.7 programs. But this is the medicine that LRSD knowingly and voluntarily decided it must take in an attempt to cure the historically low academic achievement of so many of its African-American students.

D. Ct. Order at 67.

In commenting upon LRSD's duty to ensure that a significant number of African-American students score at or above the proficient level in reading, math, and science, the district court concluded its remarks by stating, "To this end, LRSD must do what it promised to do, and what it has been ordered to do because of this promise. In the words of the poet of the Yukon, Robert Service, `a promise made is a debt unpaid.'" Id. at 68.

We add to these comments only the observation that a promise is that which has been made by the promisor, and not one that may be expanded by others beyond that which is fairly encompassed by its terms.

Nothing in what we have said in this opinion should be read as in any way relieving the Little Rock School District of its obligation to comply with the commitments it made when it entered into the settlement that culminated in the Revised Plan. Our concerns about the district court's 2002 and 2004 orders arise from the ever-heightened requirements that the District is being asked to satisfy, requirements that may seem to impose a duty of demonstrating mathematical precision at a cost and effort beyond that which the District should be required to bear.

We note that in one form or another the Little Rock School District has been under judicial tutelage for more than two decades now. We hope that it is not too much to expect that its efforts to comply with not only the requirements of the Revised Plan but also with the subsequent embellishments of those requirements will prove to be successful.

With these observations, the judgment is affirmed.

GRUENDER, Circuit Judge, concurring in part and dissenting in part.

Like the Court, I would affirm the district court's finding that LRSD was not in substantial compliance with section 2.7.1 of the Revised Plan as embodied in the 2002 Remedy. However, I respectfully dissent from the Court's judgment because I find that the district court abused its discretion in imposing the 2004 Remedy.

A consent decree "is a kind of private law, agreed to by the parties and given shape over time through interpretation by the court that entered it." Knote, 29 F.3d at 1300 (quoting Sennewald v. Univ. of Minnesota, 847 F.2d 472, 475 (8th Cir. 1988) R. Arnold, J., concurring). Although we defer in large measure to the district court's interpretation or modification of a consent decree, Knote, 29 F.3d at 1300, 1302, the district court is not at liberty to ignore what was "agreed to by the parties" by imposing new terms that lie outside the intended agreement of the parties, see, e.g., Pure Country, Inc. v. Sigma Chi Fraternity, 312 F.3d 952, 958 (8th Cir.2002) ("When construing a consent decree, courts are guided by principles of contract interpretation and, where possible, will discern the parties' intent from the unambiguous terms of the written consent decree, read as a whole."); see also Holland v. N.J. Dep't of Corr., 246 F.3d 267, 281 (3d Cir.2001) ("A court should interpret a consent decree as written and should not impose terms when the parties did not agree to those terms."); EEOC v. New York Times Co., 196 F.3d 72, 78 (2d Cir.1999) ("[A] court may not replace the terms of a consent decree with its own, no matter how much of an improvement it would make in effectuating the decree's goals." (quotation omitted)).

As the Court notes, ante at 536-537, when LRSD chose not to appeal the 2002 Remedy, the 2002 Remedy became the governing interpretation of the terms agreed to by the parties in section 2.7.1 of the Revised Plan. There is no dispute that the only hurdle remaining in LRSD's quest for unitary status is compliance with subparts A and B of the 2002 Remedy. Therefore, the district court's modification should have focused on producing compliance with those terms. Subpart A required LRSD "to assess each of the programs implemented under § 2.7" for a year and a half and "to use all of that available data and information [including other information already available apart from the assessments] in assessing the effectiveness of those programs." Subpart B required LRSD to maintain written records of the assessments containing (a) the written criteria used to assess each program, (b) the results of the assessments, including any resulting modification or elimination of a program, and (c) the names of administrators and grade-level descriptions of any teachers involved in the assessment. Also as discussed by the Court, ante at 533-534, there was a clear distinction in the 2002 Remedy between a less formal, dynamic "assessment" and a more formal, research-paper-formatted "evaluation."

Instead of focusing on enforcing compliance with the terms agreed to by the parties, however, the district court imposed terms in the 2004 Remedy that are untethered to the requirements of subparts A and B of the 2002 Remedy or section 2.7.1 of the Revised Plan. Although the district court's substitution of eight indepth "evaluations" for the agreed-upon "assessments" of each program was arguably suggested in part by LRSD's own prior attempt to substitute three broad evaluations for the individual program assessments, there is no evidence of a meeting of the minds between the parties that would allow a number of in-depth evaluations to replace the agreed-upon assessments. Therefore, the district court should have simply enforced the assessment requirement as originally set forth in subparts A and B of the 2002 Remedy. The district court's substitution of a new set of rigorous evaluations not agreed to by the parties was an abuse of discretion.

There are two other aspects of the 2004 Remedy that are even more significant abuses of discretion. First, the district court ordered LRSD to hire a new team for its Program Review and Evaluation Department ("PRE"), and the district court specified in great detail the educational background and experience that the district court required of the new team. In addition, the district court all but ordered LRSD to hire a particular expert, Dr. Ross (or a person chosen by Dr. Ross), to identify the eight "key" section 2.7 programs and prepare in-depth "step 2" evaluations of those programs. Nothing in subparts A and B of the 2002 Remedy suggested that LRSD would have to operate under such detailed personnel qualification requirements in assessing the section 2.7 programs. In settling this case, LRSD certainly did not consent to the selection of its employees and consultants by the district court.

Second, the district court introduced a requirement that LRSD's "program assessment process must be deeply embedded as a permanent part of LRSD's curriculum and instruction program" (emphasis by the district court). The district court did not identify any objective standards by which it intends to measure whether LRSD succeeds in meeting this "deeply embedded" requirement. The district court justified this requirement by stating, "Only then will I have the necessary assurance that LRSD intends to continue using that process for as long as it is needed to determine the effectiveness" of LRSD's programs in closing the achievement gap. However, the district court's responsibility is to obtain the necessary assurance that LRSD is complying with the terms of the consent decree, not to independently assess whether those terms are effective. The parties agreed to annual program assessments by administrators and teachers, not to a permanently embedded institutional structure of reviewing experts chosen by the district court.

The introduction of the impossibly subjective "deeply embedded" requirement, viewed in light of the district court's lack of restraint to date in redefining the program assessment requirements in subparts A and B and micro-managing LRSD's compliance team, raises the specter that the district court intends to retain control of LRSD's efforts to close the achievement gap regardless of whether LRSD meets the terms agreed to by the parties. No matter how much the district court believed that the new terms in the 2004 Remedy would make an "improvement ... in effectuating the decree's goal[ ]" of closing the achievement gap, a laudable motive, the district court simply "may not replace the terms of a consent decree with its own." New York Times Co., 196 F.3d at 78 (quotation omitted). For these reasons, I would find that the district court abused its discretion in imposing the 2004 Remedy.

Finally, I recognize that LRSD already has invested a substantial amount of the effort and expense needed to meet the requirements of the 2004 Remedy. At this point, vacating the 2004 Remedy in its entirety and remanding to the district court to impose a new remedy might actually set back LRSD's efforts to attain unitary status. Therefore, I would instruct the district court to order LRSD to complete the eight "step 2" evaluations as called for in the 2004 Remedy, but I would also instruct the district court to analyze those eight evaluations under the standards set forth in subparts A and B of the 2002 Remedy. Specifically, the evaluation reports individually would need to meet the standards of a "program assessment," rather than an "evaluation," as defined ante at 533-534; LRSD would have to show that it used "all of th[e] available data and information in assessing the effectiveness of those programs" as required by subpart A, meaning that the in-depth data gathered in the "step 2" evaluations would be put to good use; and continued compliance with the written record requirements specified in subpart B of the 2002 Remedy would supplant the unworkably subjective "deeply embedded" standard created out of whole cloth in the 2004 Remedy. The detailed personnel requirements imposed on LRSD would be vacated. Such a modification of the 2004 Remedy would restore the standards agreed to by the parties without disrupting the substantial compliance efforts LRSD has undertaken to date with respect to that remedy.

Accordingly, I respectfully dissent from the Court's decision to uphold the 2004 Remedy and would modify the 2004 Remedy as discussed.