For more than thirty years after Brown the elementary schools in Hattiesburg, Mis*386sissippi, have remained almost totally segregated. In 1970 the United States filed suit against the State of Mississippi and several State agencies and officials, seeking the desegregation of thirteen school districts. The Hattiesburg Municipal Separate School District (HMSSD) intervened as a defendant. The district court approved a pupil assignment plan embodied in a consent decree between the United States and HMSSD and in 1971 approved the school system’s plan for further student desegregation, but the case lay dormant for a dozen years. The HMSSD filed the reports that it was required to file by the 1970 and 1971 orders. The district court found, however, “That the HMSSD has complied with all previous orders does not render the system unitary.” In February 1984, certain black children intervened, alleging that the HMSSD elementary schools 1 had never been adequately desegregated and that the United States had failed to protect the interests of black children in the district.
I. Background
HMSSD presents fewer barriers to desegregation than most school districts. Hattiesburg has a population of 40,000. The northeast and south residential areas of the city are predominantly black; the central and west areas of the city are predominantly white. The residential areas in the east and east-central areas of the city are more integrated. The school district is small, compact, and logistically manageable from the standpoint of desegregating by the pairing and clustering of schools. In its March 1985 report the HMSSD showed that there were in elementary schools, in grades one through six, 1740 blacks (59%) and 1195 whites (41%). Desegregation of secondary schools2 seems to have been adequate and proceeded without turmoil.
Five of eleven of Hattiesburg’s elementary schools are virtually all-black; three are more than 77 percent white, containing 905 of the 1173 white pupils. More than one third of all black children (609) attend Bethune School; five whites attend. Excluding students in grades other than 1-6 and special education students, at Grace Love School 100 percent of the children are black; at Jones 99 percent; at Eureka 94 percent; at Walthall 91 percent. Almost half of all whites in the elementary grades, 526, constitute 90 percent of the white pupils at Thames. At Grace Christian 80 percent of the children are white. At Woodley, 72 percent are white.
The HMSSD’s March 15, 1985 report filed with the court shows enrollment by race in the elementary schools:
School Blacks Whites Total % Black
Bethune 609 5 614 99
Jones 191 24 215 89
Eureka 165 12 177 93
Love 138 -0-138 100
Walthall 172 17 189 91
Eaton 57 38 95 60
Davis 116 66 182 64
Camp 101 109 210 48
Christian 47 184 231 20
Thames 57 526 583 10
Woodley87 214 301 29
Total 1740 1195 2935 59
*387These figures show a somewhat smaller differential between blacks and whites than the plaintiffs’ figures because they include students not in grades 1-6 (pre-K and kindergarten) and also special education students.
In July 1984 the HMSSD and the United States entered into and filed with the court a proposed consent decree providing for the modification of attendance zone lines, creation of two magnet schools, increase of majority-to-minority transfers, and educational improvements of historically black schools. The court did not rule on this proposed consent decree.
In September 1984 the court approved a consent decree submitted by the HMSSD and the plaintiffs-intervenors setting out a procedure for the submission of alternative desegregation plans for the district elementary schools. Under that decree, at the request of HMSSD, the Title IV Racial Desegregation Assistance Center,3 and Dr. Larry Winecoff of the University of South Carolina, assisted by Dr. Barnett Joiner of Grambling University, submitted two plans, referred to as Foster Plan A4 and Foster Plan B.5 The HMSSD submitted its District Plan of December 10 6 and District Alternative Plan of December 10.7 The Superintendent’s Biracial Committee8 approved both plans. On January 21, 1985, the plaintiffs-intervenors filed a plan prepared by Dr. Michael Stolee of the University of Wisconsin-Milwaukee, based on pairing and clustering schools. The United States employed Dr. Christine Rossell, a political scientist at Boston University, to evaluate the plans in terms of interracial exposure produced.9 Dr. Rossell chose the District Alternative Plan as the best plan. The United States, the HMSSD, and the State of Mississippi entered into a proposed consent decree substantially the same as the District Alternative Plan with most of the changes Dr. Rossell recommended.
II. The Consent Decree or Magnet Plan
This plan, approved by the district court, is based on the conversion of two of the five virtually all-black elementary schools (Jones and Walthall) into magnet facilities, with specialized curricula. Equal numbers of black and white students to the number 240 would be admitted to these schools upon approval of their voluntary applications, so as to maintain a 50 percent black, 50 percent white enrollment in each facility. A third black school (Grace Love) would be consolidated with a small, racially *388mixed facility (Eaton) that would be closed, resulting in a projected enrollment in grades one through six that would be 73 percent black. The two remaining virtually all-black schools, Bethune and Eureka, would neither become “magnets” nor have any white students in grades 1-6 reassigned to them. Up to 188 black students would be transferred from Walthall. Finally, several attendance zone changes would be made that would reassign black children to formerly white schools.
This appeal turns on the adequacy of the Consent Decree Plan to desegregate the elementary grades in Hattiesburg. In the opinion of this Court that plan is like a voice from the past crying for “gradualism”. Constitutionally, its desegregative effect is too little and too late. All of the parties seem to have worked in good faith, with wide spread community support, and with the help of experienced experts. Some of the educational improvements seem highly desirable, but do not necessarily bear any relation to desegregation. We remand the case to the district court with the suggestion that the parties try again and that they accelerate the desegregation process.
III. Magnet Schools and Effect on Other Schools
We start with the fact that until now the HMSSD has made no serious effort to desegregate its elementary grades.10 This long-continued operation of the dual system calls for its prompt and effective dismantling now.
The action the HMSSD proposes to take, with the district court’s blessing, is primarily to provide two magnet schools, Jones and Walthall. These will not become operative until the school year, 1987-88. Whether successful or not, they will continue to function for three years. The decree was rendered in 1985; not until the end of 1991, therefore, will the court evaluate the effectiveness of the magnet schools. That is because all the experts agree that three or more years are needed for the program to become effective. For years and years this Court and the Supreme Court, in too many decisions to cite, have said the time is past for courts to postpone adequate deseg1 regation of schools.
Each of the schools is projected to enroll at its peak 240 students. At most, therefore, 240 blacks will be desegregated by the magnet plan. Dr. Rossell conceded, however, that the school district “does not expect” Jones and Walthall “to be up to capacity.” 11 As Dr. Rossell testified, reading from onS of her published papers, “Schools located in minority neighborhoods tend to be under-enrolled in by whites.”12 Jones and Walthall, now 89 percent and 91 *389percent black, respectively, are located in black neighborhoods. If only 30 whites apply to Jones or Walthall, that school would be limited to an enrollment of 60.
First, we make the general observation that to the extent the magnet schools draw black pupils, the number of blacks in identifiably black schools will increase. The HMSSD estimates that up to 188 blacks now in Walthall will be added to Eureka, increasing its black population from 93 percent to 97 percent. We turn to the other schools. Grace Love with 138 students, all black, would be consolidated with Eaton, a small school with 57 blacks and 38 whites, which would be closed. Grace Love would receive 38 whites, who would be more segregated. Bethune will continue to be be 99 percent black although there would be some reassignment because of rezoning. Thus, 62 blacks would be rezoned to Woodley, now only 29 percent black, and 50 would be rezoned to Thames, now only 10 percent black.
There is no question at all about Eureka; it is expected to be a virtually all-black school for the foreseeable future. Dr. Spinks, HMSSD Superintendent from 1966 to 1985 testified: “I just don’t think that they’ll [whites will] go to that school.” Tr. 88. In fact, under the proposed plan, because of reassignments required by conversion of Walthall to a magnet school, the school’s enrollment will have about twice as many black pupils as it now does.
The HMSSD’s experts, as well as Dr. Stolee for the plaintiffs, agreed that Bethune would remain almost totally black. Any expectation that white students would voluntarily transfer to Bethune seems unrealistic in view of the last twenty-one years, when no white elementary student exercised his freedom of choice or majority-to-minority “right” to attend Bethune or any other black school. Bethune, Eureka, and Grace Love have the lowest test scores in the system. It seems unlikely, therefore, that white students would attend these schools to take advantage of “Basic Skills Learning Centers” to be established. Restricting the Extended Day program to Bethune with the idea of attracting a hundred white students across grades 1-6 seems equally illusory.
The following table, offered by the plaintiffs but apparently accurate, compares the attendance changes in grades 1-6 of the Magnet or Consent Decree Plan with the Stolee Plan for pairing and clustering schools:
MAGNET PLAN STOLEE PLAN 1986-87
% Black Projected Changes in Projected
School March 1985 % Black Attendance % Black
Bethune 99% 99% Clustered with 61% Thames and Love
Eureka 93% 97% Clustered with 63% Jones and Christian
Grace Love 100% 73% Clustered with 54% Bethune and Thames
Jones 89% 50% Clustered with 65% Christian and Eureka
Walthall 91% 50% Paired with 53% Woodley
Eaton 60% Remains open or 54% included in Thames/ Bethune/Love or (if Jones/Christian/ open) Eureka clusters
*390MAGNET PLAN STOLEE PLAN 1986-87
Jo Black Projected Changes in Projected
School March 1985 % Black Attendance_% Black
Davis 64% 64%
Camp 48% 47%
Woodley 29% Paired with Walthall 53%
Grace Christian 20% Clustered with Jones and Eureka 64%
Thames 10% Clustered with Bethune and Love 59%
We do not condemn magnet schools, as such. We say that to rest the desegregation program on the magnet plan with only other minor adjustments does not comply with the Supreme Court’s mandate in Swann v. Charlotte-Mecklenburg: There must be “every effort to achieve the greatest possible degree of actual desegregation”.13 The plan must be one that “promises realistically to work and promises realistically to work now.” Green v. Country School Board.14
The defects in the Consent Decree or Magnet Plan are these:
(1) The operation of magnet schools is postponed until the school year 1987-88;
(2) The evaluation of the merits of the schools is deferred until 1991;
(3) The student population in the magnet schools is so small and the chance of achieving even the limited goal of 240 students in each magnet school is so slight that the plan fails to achieve adequate desegregation of the system as a whole;
(4) Bethune and Eureka, the two most racially isolated black schools will be scarcely affected by the magnet schools or the plan as a whole.
We see no escape from the conclusion that the magnet school plan, accompanied only by the small adjustments contemplated, do not meet the constitutional test for dismantling a long established dual system. Magnet schools should be a supplement to a mandatory desegregation plan based to a reasonable extent on mandatory reassignment and pairing and clustering of schools.
IV. White Flight
As the district judge pointed out in his opinion: “White flight” is, of course, no justification for inaction. See United States v. Scotland Neck City Board of Education, 407 U.S. 484, 92 S.Ct. 2214, 33 L.Ed.2d 75 (1972). A court may, however, “consider anticipated white flight and choose the desegregation plan most likely to minimize white boycotts”. The district court held that the “Consent Decree Plan will cause less white flight ... all the while further desegregating the system and increasing interracial exposure”; on the other hand, the “Stolee Plan is much less likely to achieve the required result of further desegregation”. “While initial projections of its success appear promising ... the Stolee Plan will not ultimately lead to more fully desegregated elementary schools in the HMSSD.” 622 F.Supp. 622.
We hold that the district court was clearly erroneous. This case does not present a choice between two constitutional plans. The Consent Decree Plan, even if the magnet schools come up to expectations, is fatally flawed, because it leaves virtually untouched two historically black schools, Bethune and Eureka, each having almost a totally black student body. We have here not a large number of schools with a few segregated schools. Bethune and Eureka, two of five segregated schools, have 40 *391percent of the black students in the elementary grades.
It is obvious that the Consent Decree was constructed with white flight as its premise. By keeping Bethune and Eureka segregated and by placing a low ceiling of 120 black students in each of the magnet schools “to make those magnet schools more attractive to the white community”,15 the Consent Decree Plan hoped to reduce white flight. This concern, legitimate when choosing among constitutionally permissible plans, cannot be “accepted as a reason for achieving less than complete uprooting of the dual public school system”.16 “White flight must be met with creativity, not with delay in segregation.” 17.
Dr. Rossell, the government’s star witness, has excellent credentials and has specialized in what is termed interracial exposure, that is, the “percentage of white students in the average Black child’s school”.18 She testified that after two years of the magnet school plan, it would produce more interracial contacts than the Stolee plan. This may or may not be true, depending on the degree to which whites will send their children to the magnet schools and the extent of white flight from other schools. On cross-examination, Dr. Rossell testified that her calculations were not exact enough to establish that the difference between the level of “interracial exposure” under the Stolee plan and the magnet plan was more than 0.2 percentage points.19 Perhaps because of this slight difference, the district court did not rely on Dr. Rossell’s interracial exposure thesis, much as it occupied importance in the trial.20 The district court relied squarely on Dr. Rossell’s prediction that the Stolee plan would cause greater white flight than the magnet plan.21
Past history, future projections of school attendance, and the continued existence of the historically all-black schools of Bethune and Eureka undermine the trial court’s conclusionary finding.
V. Conclusion
We are by no means persuaded that Dr. Stolee’s plan, as proposed by the plaintiffsintervenors, will necessarily solve Hatties*392burg’s elementary school problems. A close study of the record, however, convinces us that the Consent Decree plan offers too little and flies in the face of the need for more action now. There are some good things in that plan. The educational improvements are excellent and should be carried out. But those educational improvements have no necessary relation to desegregating schools.22 Magnet schools are, in general, good for a school system. But magnet schools cannot be used, as they appear to be used here, for example, partly as a sop to Cerberus, to attract white support and, hopefully, deter white flight. Some mandatory assignment and some pairing and clustering of schools will be necessary to uproot the historically segregated dual system of elementary schools in Hattiesburg.
Now is the time to move. The secondary school system seems to be desegregated and operating adequately. The HMSSD, with the aid of the United States, has been able to muster community support for a major move in the direction of desegregating the elementary grades. The Superintendent’s Biracial Committee represents a broad cross-section of the community. The district judge instituted a model system of soliciting desegregation plans from qualified experts. His efforts and his thorough knowledge of the various plans, as shown in his thoughtful opinion, make us reluctant to take any radical step. We have concluded therefore, that we should remand this case to the district court with the suggestion that it direct the parties to accelerate desegregation and consider mandatory reassignment, some pairing and clustering of schools, supplemented by magnet schools. We particularly find it objectionable that Bethune and Eureka should remain, as they have always been, almost totally black.23
The judgment of the district court is reversed and remanded for further proceedings consistent with this opinion.
. The secondary school system is not at issue in this litigation. After entry of the 1971 court order the secondary schools were fully desegregated through pairing and grade consolidation.
At the secondary level, the school system enrolled 1158 (45%) white and 1411 (55%) black students. Individual school enrollments were as follows:
Black % White %
Burney (grade 9) 140 (52) 130 (48)
Hawkins (7-9) 365 (70) 155 (30)
Thames Jr. High (7-8) 303 (52) 282 (48)
Rowan (10) 238 (56) 189 (44)
Blair (11-12) 365 (48) 402 (52)
. See footnote 1.
. The center, funded under Titte IV of the Civil Rights Act of 1964, serves Mississippi.
. Foster Plan A is a magnet school plan with a partial same-year mandatory reassignment backup plan. It proposes the use of Bethune, Jones, and Walthall as magnet schools, reassigning students currently attending those schools to predominantly white schools. The plan also recommends closing three schools, two of which are historically black.
. Foster Plan B is a mandatory reassignment plan. It proposes the closing of two historically black schools. Eureka and Love, and the establishment of extended day programs at one formerly black school and one formerly white school.
. The District Plan is similar to the Consent Decree Plan submitted in 1984 by the United States and HMSSD.
. The District Alternative Plan adopts parts of the other plans submitted. With the addition of some of the changes recommended in Rossell’s report, it comprises the Consent Decree Plan before the court.
. The Superintendent’s Biracial Committee is composed of equal numbers of black and white residents of the district, including parents and non-parents who live in different attendance zones. The Committee advises the Superintendent on various issues regarding the school district and also actively participated in planning and drafting the desegregation plans submitted by the HMSSD. At trial, several members of the Committee testified, including Charles Lawrence, an attorney, newly elected member of the Hattiesburg City Council and Chairman of the NAACP Education Committee.
. “Interracial exposure” is the percentage of white students in the average black child’s school. The trial judge stated that it "measures net benefit more accurately than a consideration of only racial balance”. Foster Plan A would produce a greater degree of interracial exposure than the Consent Decree Plan. Dr. Rossell testified that the Foster Plan A is not the most effective because it would close three schools, two of which are historically black. This is not favored by the Biracial Committee.
. In 1964 the HMSSD used a freedom-of-choice plan under which no white student ever chose to attend a black school. The 1970 decree used pairing and grade consolidation but it was based on a system of contiguous geographic zoning which did not alter the historic racial identifiability of the schools. See Table of Enrollment by Schools. Although majority-to-minority transfers are authorized, no white student even applied for transfer, blacks averaged about 50-60 annually in recent years.
The government’s expert witness. Dr. Rossell, recognized that under the magnet plan, hundreds of Black students would remain in racially identifiable Black schools for years — but she supported this feature of the plan on the theory that the Black students’ failure to exercise majority-to-minority transfers made these schools constitutionally acceptable “by choice”. Tr. 611. Dr. Rossell has excellent credentials as an expert on “interracial exposure", but this Court, with long experience in the field of school desegregation, knows that burdening black parents with the obligation of choosing schools is unworkable in fact and contrary to the law. The district school board not black parents and their children bears the burden of desegregating schools. Green v. County School Bd. of New Kent County, 391 U.S. 430, 441-42, 88 S.Ct. 1689, 1696, 20 L.Ed.2d 716 (1968); Raney v. Board of Educ. of Gould, 391 U.S. 443, 447-48, 88 S.Ct. 1697, 1699-1700, 20 L.Ed.2d 727 (1968); Monroe v. Board of Comm’rs of Jackson, 391 U.S. 450, 458, 88 S.Ct. 1700, 1704, 20 L.Ed.2d 733 (1968).
. Tr. 602.
. Continuing, she read: "This was the case in Milwaukee where the inner-city magnets were 40 percent empty and in Boston where the only magnet school to fail to meet its projected enrollment was the Martin Luther King Middle School in the heart of Roxbury, a primarily black community.” Tr. 602.
. 402 U.S. 1, 26, 91 S.Ct. 1267, 1281, 28 L.Ed.2d 554 (1971).
. Green v. School Board of New Kent County, 391 U.S. 430, 439, 88 S.Ct. 1689, 1694, 20 L.Ed.2d 716 (1968).
. Testimony of HMSSD’s Dr. Walker. There was disagreement about the extent to which the magnet plan and the Stolee plan would distribute the burdens of desegregation equitably. Dr. Stolee estimated the total number of students subject to mandatory reassignments. Under the Stolee plan, 604 black and 582 white pupils will be reassigned to different schools for three, or possibly four, of the six elementary grades (Tr. 684; see Tr. 211-12, 669-72, 780). Superintendent Walker agreed that these figures were relatively equitable (Tr. 824). In contrast, under the magnet plan’s attendance changes far more black than white students would be reassigned, and for all of the elementary grades. "[T]he 497 black children involuntarily transferred compared to the 65 white children is ... inequitable.” Tr. 686.
Dr. Walker did not view many of these mandatory reassignments as burdensome because "[tjhere are certain students in the School District plan which would be reassigned to a school which should be or which ought to be their neighborhood school” (Tr. 820); these reassignments, he said, are a “[djisruption ... [but] not a burden" (Tr. 834; accord, Tr. 581 [Dr. Rossell]; but see Tr. 586 [individual students and parents might take a different view]). Excluding such changes. Dr. Walker calculated the magnet plan’s "burdensome reassignments” as affecting 178 black and 0 white students. Tr. 820-24.
. United States v. Scotland Neck City Board of Education, 407 U.S. 484, 491, 92 S.Ct. 2214, 2218, 33 L.Ed.2d 75 (1972).
. Davis v. East Baton Rouge Parish School Bd., 721 F.2d 1425, 1438 (5th Cir.1983).
. Tr. 550.
. Tr. 622-23.
. "[T]he testimony showed that the difference between the initial net benefit and effectiveness of the Consent Decree Plan and that of the Stolee Plan would be slight”. See R. 714, R.Esc. 74 [Mem.Op. 12, text at n. 28]. The Stolee plan "has interracial exposure of only a few percentage points lower than that of the Consent Decree Plan”. R. 716 R.Exc. 76 [Mem.Op. 14].
. ”[T]his court is of the opinion, based on the testimony of Rossell and other experts, that the Stolee Plan will not ultimately lead to more fully desegregated elementary schools in the HMSSD" (emphasis supplied). Tr. 717, R.Exc. 77.
. Dr. Stolee testified:
Now, they’ve got some good ideas. They’ve got the basic fundamentals program that they were talking about, the basic skills program that they propose to put at, I believe, Bethune, Eureka and Love; and I commend them for doing that____ But I don’t think they need desegregation as an excuse to do it. The presence or absence of a desegregation plan does not mean that they can or cannot add these good programs, and I’m assuming they’ll do it.
Secondly, I think they’ll come up with some very good ideas for magnet schools. I have to say I think so, because as yet they’ve not told us what they’re going to do with their magnets except the process, and the process is good. And those magnet schools can help provide better education, but it’s also possible to put in magnet schools as a part of a mandatory assignment school desegregation plan.
I know I’ve been going on at length, but the main point is that every single educational improvement that the school system wants to and I know has in the past can be made functional within a mandatory reassignment plan.
[TJhat [magnet] plan, I think, is a good educational plan. But this case has to do with desegregation, and it does not desegregate the School District. And I maintain that your fine educational plans can be put in place with the plaintiff-intervenors’ plan which does desegregate the School District.
. Davis v. East Baton Rouge Parish School Board, 721 F.2d 1425, 1436 (5th Cir.1983).
. United States v. Lawrence County School Dist., 799 F.2d 1031 (5th Cir.1986).