The United States and various black and Mexican-American intervenors have chai-, lenged the student assignment policies of the Austin Independent School District (AISD). This is the second time this case has come before us. In 1972, our en banc Court remanded the case to the district court with directions to eliminate all discriminatory segregation against black and Mexican-American students and to establish a unitary school system in Austin. United States v. Texas Education Agency, 5 Cir., 467 F.2d 848 (Austin I). At the time of that decision, the AISD was 65 percent Anglo, 20 percent Mexican-American, and 15 percent black. Eighty-three percent of the black students and 58 percent of the Mexican-Americans attended schools that contained over three-fifths minority1 students. The district court, on remand from our en banc decision, adopted the desegregation plan submitted by the AISD. This plan has had two years of operation to prove itself. The school system is now 62 percent Anglo, 23 percent Mexican-American, and 15 percent black. Forty-two percent of the black Austin students and 45 percent of the Mexican-Americans still attend schools that are over three-fifths minority. Progress has been made. But the AISD is far from a unitary system.
This Austin case differs from the one we considered in 1972 in two respects. First, we must weigh the effect of the Supreme Court decision in Keyes2 on the burdens of the plaintiffs and defendants. Second, we must measure the constitutional sufficiency of the new desegregation plans the AISD and the intervenors have submitted.
I.
PROCEDURAL HISTORY
This school desegregation case was filed in August 1970 by the United States against the Texas Education Agency and seven school districts, including the AISD. The complaint alleged that (1) historically, the defendants had operated a dual system based on race, and continued to do so, and (2) the defendants discriminatorily assigned Mexican-Americans to schools identifiable as Mexican-American schools or as schools intended for blacks and Mexican-Americans. Certain blacks and Mexican-Americans intervened on their own behalf and as representatives of those similarly situated.
After the parties and the United States Department of Health, Education, and Welfare were unable to agree on a desegregation plan, the district court consolidated a hearing that took place June 14 to June 21, 1971. The court held that there had been no de jure discrimination against Mexican-Americans and afforded them no relief. It then held that the “vestiges of a dual system continue to exist with respect to blacks” and adopted, with minor modifications, the AISD plan for establishing a unitary school system in Austin. The high schools and junior high schools were to be desegregated primarily by busing about 2200 blacks to previously predominantly *385white schools. The elementary schools were “clustered” into groups of six schools each. One week per month the students of each cluster were to meet together to engage in certain planned activities. The district court found “that elementary students would be in a desegregated environment as much as twenty-five (25) percent of the school year”.
This Court, sitting en banc, reversed and remanded the case to the district court with directions to eliminate the unconstitutional segregation of Mexican-American and black students “at once”. 467 F.2d at 883. We held that the AISD had caused and perpetuated the segregation of blacks within the school system and that it had not dismantled this dual system. The Court further held that the educational status of Mexican-American students was inferior to that of their Anglo counterparts and that Mexican-Americans in Austin were a separate ethnic minority within the ambit of the Equal Protection Clause. Because school authorities, by their actions, contributed to the segregation of Mexican-Americans in, the Austin schools, we held that these students were denied the equal protection of the laws.
The en banc Court divided only on the issue of remedy. A majority of the Court directed the district court to eliminate the dual school system and itemized a hierarchy of desegregation tools that the court should consider using. Six judges dissented: “The majority opinion . . . [is] an example of how a reviewing court can pass the buck, give the school board a delay, and confuse the district court on remand.” 467 F.2d at 888. This evaluation has to some extent been borne- out by later events. The district judge admitted to the attorneys in this case that he was baffled by the majority opinion on remedy and asked for help,in interpreting it. The response of the attorneys (of both the intervening plaintiffs and the AISD) was to move for clarification of this Court’s mandate. The motion was denied over'the dissent of five judges. United States v. Texas Education Agency, 5 Cir. 1973, 470 F.2d 1001 (en banc). .
On August 3, 1972, the day after our en banc decision was issued, the district court ordered the parties to hold a pretrial conference within five days to discuss the possibility of joining in the submission of a single desegregation plan to that court. If no agreement could be reached, the AISD, the United States, and the intervenors were ordered to submit plans by August 14. The United States reported to the district court on that day that it was “unable to submit a desegregation plan at this time” and recommended to the court “that local officials be given the opportunity to formulate and submit a plan to the Court before the Court or other parties consider alternatives or modifications to such a plan”. On the same day, the intervenors and the AISD filed desegregation plans. The government has yet to file any plan.
The AISD’s plan would establish six sixth grade centers that would draw all sixth-graders in the school district. The interve-nors’ plan would require the busing of all kindergarten-to-fourth-grade students (K-4) at the predominantly minority schools in East Austin to- new grade K-4 schools in West Austin, and the busing of all fifth-to-eighth-grade students at predominantly Anglo schools in West Austin to new grade 5-8 schools in East Austin. The plan would also close the one predominantly minority high school remaining in Austin (Johnston) and bus its students to the remaining high schools. The black intervenors added an objection to the 1971 closing of the black high school- (Anderson) and black junior high school (Kealing) in Austin, and requested that the schools be reopened and used in any desegregation plan adopted by the court. The district judge conducted the trial for twelve days in May, 1973. He issued a “Memorandum Opinion and Order” on August 1, 1973.
The district court first held that, because the AISD had, in the past, intentionally segregated black students, it must now dismantle its dual system based on race. Second, the court held that its finding of past intentional segregation of blacks constituted a prima facie case of intentional *386segregation of Mexican-Americans. It concluded, however, that the AISD had successfully rebutted this prima facie case by demonstrating that its racial policies had been unrelated to its treatment of Mexican-Americans and that there was an absence of segregative intent toward Mexican-Americans. The court, relying on Keyes v. School District No. 1, Denver, Colorado, 1973, 413 U.S. 189, 93 S.Ct. 2686, 37 L.Ed.2d 548, held that it would therefore be improper to order “all-out desegregation” of Mexican-Americans. The court then rejected the black intervenors’ challenge of the closing of Anderson and Kealing schools and adopted, with minor modifications, the AISD’s plan for establishing an integrated school system. Because Mexican-American students are an identifiable minority entitled to equal protection of the laws, the court further held that they were entitled to equal educational opportunities, including the setting up of special educational programs, such as bilingual and bicultural education.
The United States and the plaintiff-inter-venors have appealed from this Memorandum Opinion and Order of the district court.
II.
SEGREGATION OF MEXICAN-AMERICANS
A. The Keyes Case
The Supreme Court held in Brown v. Board of Education, 1954, 347 U.S. 483, 495, 74 S.Ct. 686, 692, 98 L.Ed. 873, 881, that educational facilities segregated on the basis of race are inherently unequal. In Keyes v. School District No. 1, Denver, Colorado, 1973, 413 U.S. 189, 195-98, 93 S.Ct. 2686, 2690-2692, 37 L.Ed.2d 548, 555-557, the Court extended this principle to the segregation of Mexican-Americans in the Denver school system. The unequal educational status of these minorities does not constitute a violation of the Equal Protection Clause of the Fourteenth Amendment unless it results from “state action”. The term of art that has long described the state action requirement in the school desegregation context is “de jure segregation”, which the Supreme Court has defined as “a current condition of segregation resulting from intentional state action directed specifically to the [segregated] schools”. Keyes, 413 U.S. at 205-06, 93 S.Ct. at 2697, 37 L.Ed.2d at 561. See generally Cisneros v. Corpus Christi Independent School District, 5 Cir. 1972, 467 F.2d 142, 148 (en banc), cert. denied, 1973, 413 U.S. 920, 93 S.Ct. 3053, 37 L.Ed.2d 1041. To establish a prima facie case of unlawful school segregation, the plaintiffs must therefore prove (1) that there is segregation in public schools, (2) that state officials have, with segregative intent, taken or refrained from taking certain actions, and (3) that the present segregated system is a result of that action or inaction.3
The remainder of the Keyes opinion considers whether proof of de jure segregation *387in a portion of the Denver school district is sufficient to establish a system-wide constitutional violation. This section of Keyes is irrelevant to our disposition of the case before us because we hold below that the plaintiffs have proved that intentional segregation exists throughout the Austin school district.
B. The Gisneros-Austin I Test
We found in Austin I that Mexican-Ameriean students in Austin had received an education inferior to that of their Anglo counterparts and that this was the result of ethnic segregation. 467 F.2d at 862-63 & n.21. This would constitute an equal protection violation, we held, only if the “school authorities, by their actions, [had] contribute^] to segregation in education, whether by causing additional segregation or maintaining existing segregation . . . .” 467 F.2d at 863-64. Our ultimate decision against the AISD was based in part on our finding that “[t]he natural and foreseeable consequence of [its] actions was segregation of Mexican-Americans”. 467 F.2d at 863. We held, however, that, to establish an equal protection violation, it is not necessary to prove discriminatory intent when there is discriminatory effect. 467 F.2d at 864-65 n.25.
Although, in Cisneros, we discarded “the anodyne dichotomy of classical de facto and de jure segregation”, the rationale of that decision was very similar to that of Austin I. The Court held that, in order to sustain a constitutional violation,
[we] need only find a real and significant relationship, in terms of cause and effect, between state action and the denial of educational opportunity occasioned by the racial and ■' ethnic separation of public school students.
Id. As in Austin I, we held that “[discriminatory motive and purpose . . . are not necessary ingredients of constitutional violations in the field of public education”. 467 F.2d at 149. And, in language reminiscent of the Austin I “foreseeable consequences” approach, the Court found the requisite state action in Cisneros in the Board’s imposition of “a neighborhood school plan, ab initio, upon a clear and established pattern of residential segregation in the face of an obvious and inevitable result”. Id.
Thus, Austin I and Cisneros both applied cause-and-effect tests for finding the state action that is a prerequisite to establishing a constitutional violation. But, in both cases, this test was applied in the context of school board actions that led to the “foreseeable” and “inevitable” result of segregated schools.
C. The Impact of Keyes on the Cisneros-Austin I Test
The Mexican-American intervenors argue that the cause-and-effect test need not fall by the wayside after Keyes because that case does not establish that segregative intent is a necessary element of unconstitutional school segregation. The intervenors point out that the Keyes holding is limited by the plaintiffs’ concession in that case that they had the burden of proving intentional state action and by the obvious segre-gative purpose of the Denver school authorities. See generally Hart v. Community School Board of Education, New York School District # 21, 2 Cir. 1975, 512 F.2d 37, 49; Comment, Public School Segregation and the Contours of Unconstitutionality: The Denver School Board Case, 45 Colo. L.Rev. 457, 475 (1974). This Court has already rejected this argument. In Morales v. Shannon, 5 Cir. 1975, 516 F.2d 411, 412-13, cert. denied, 1975, 423 U.S. 1034, 96 S.Ct. 566, 46 L.Ed.2d 408, we held:
[W]ith respect to the first issue, segrega-tory intent, we are governed by Keyes . , which supervened our holding in Cisneros . . . , to the extent that Keyes requires, as a prerequisite to a decree to desegregate a de facto system, . proof of segregatory intent as a part of state action.
Morales also compels rejection of the in-tervenors’ argument that the Cisneros-Aus-*388tin I test for the constitutional violation is the “functional equivalent” of the Keyes test. To the extent that Cisneros and Austin I applied cause-and-effect tests and rejected the requirement of a showing of discriminatory intent, those cases were supervened by Keyes.
But the intervenors also argue that, although this Court in Cisneros and Austin I refused to search for the defendants’ express or specific intent, we did not discard intent as an element of the equal protection violation. The intervenors contend that intent could be inferred in those cases from our findings that segregation was the “inevitable result” and the “foreseeable consequence” of the school boards’ actions. Whatever may have been the originally intended meaning of the tests we applied in Cisneros and Austin I, we agree with the intervenors that, after Keyes, our two opinions must be viewed as incorporating in school segregation law the ordinary rule of tort law that a person intends the natural and foreseeable consequences of his actions.4 This reading of Cisneros and Austin I is faithful to the Keyes requirement of proof of segregative intent. See Comment, 45 Colo.L.Rev. at 464 (1974). But see Comment, Keyes v. School District No. 1: Unlocking the Northern Schoolhouse Doors, 9 Harv.Civ.Rights-Civ.Lib.L.Rev. 124, 149 n.99 (1974).
Apart from the need to conform Cisneros and Austin I to the supervening Keyes case, there are other reasons for attributing responsibility to a state official who should reasonably foresee the segregative effects of his actions. First, it is difficult—and often futile—to obtain direct evidence of the official’s intentions. Rather than announce his intention of violating antidis-crimination laws, it is far more likely that the state official “will pursue his discriminatory practices in ways that are devious, by methods subtle and illusive—for we deal with an area in which ‘subleties of conduct . play no small part’ ”. Holland v. Edwards, 1954, 307 N.Y. 38, 45, 119 N.E.2d 581, 584. See also United States v. O’Brien, 1968, 391 U.S. 367, 383-85, 88 S.Ct. 1673, 1682-1683, 20 L.Ed.2d 672, 683-684; Brest, Palmer v. Thompson: An Approach to the Problem of Unconstitutional Legislative Motive, 1971 Sup.Ct.Rev. 95, 124. Hence, courts usually rely on circumstantial evidence to ascertain the decisionmakers’ motivations.5
*389Second, in Monroe v. Pape, 1961, 365 U.S. 167, 187, 81 S.Ct. 473, 484, 5 L.Ed.2d 492, 505, the Supreme Court rejected the argument that specific intent is a necessary-element of the cause of action under 42 U.S.C. § 1983, the statute under which many school desegregation eases are brought. The Court held that § 1983 “should be read against the background of tort liability that makes a man responsible for the natural consequences of his actions”. Id. We find no inconsistency between the rule applied in Monroe v. Pape and that applied in Keyes, nor do we find any reason for applying a standard different from Monroe v. Pape in school desegregation cases.6 See Ely, Legislative and Administrative Motivation in Constitutional Law, 79 Yale L.J. 1205, 1292 n.258 (1970).
One final word need be added about our Austin I formulation. Our holding that there was unlawful state-imposed segregation was based in part on our finding that affirmative action by the school authorities could have resulted in desegregation. 467 F.2d at 863. The AISD criticizes this approach because it would put
virtually all school districts under massive desegregation orders. Racial and ethnic imbalances occur wherever there are racial or ethnic minorities, and no school district can measure up to a standard which requires that it have taken affirmative action to promote the integration of all racial and ethnic minorities throughout its history.
Our holding in Austin I placed no such burden on school boards. Our statement about affirmative action immediately followed our finding that the foreseeable consequence of various actions of the AISD was the segregation of Mexican-Americans. Hence, our holding of unlawful segregation was based on the foreseeability and avoida-bility of that segregation. See Fiss, The Charlotte-Mecklenburg Case—Its Significance for Northern School Desegregation, 38 U.Chi.L.Rev. 697, 706 (1971). Many circuits have taken this approach.7 In any event, it should be clear after Keyes that the refusal of school authorities to take affirmative action that would desegregate the school system may be probative of the segregative intent underlying various actions of those officials.8
D. The Prima Facie Case of Unlawful Segregation of Mexican-Americans in Austin
1. Segregation in the schools. The district court found that there was substantial segregation of Mexican-Americans in the Austin school system. That finding is not clearly erroneous. Our Court has held that “[ujnder Keyes . . . and Cisne-ros . . ., schools in Texas with a combined predominance of black and Mexican-American students are eligible to be classi*390fied as ‘segregated schools’.”9 The statistics paint a clear picture of the extensive segregation that still exists in the Austin schools.10 Of the 41,174 students attending one of the 70 elementary and junior high schools in Austin, 16 percent (6590) are black, 24 percent (9950) are Mexican-American, and 60 percent (24,634) are Anglo. About 52 percent (3396) of the black pre-high school students and over 54 percent (5380) of the Mexican-American pre-high school students attend one of 18 schools that is over three-fifths minority. Over 47 percent (11,610) of the Anglo pre-high school students in Austin attend one of the 24 schools that is over four-fifths Anglo.11 Of the 17,746 public high school students in Austin, 14 percent (2520) are black, 19 percent (3316) are Mexican-American, and 67 percent (11,910) are Anglo. About 17 percent (423) of the black high school students and over 30 percent (1003) of the Mexican-American students attend Johnston High School, which is 99 percent minority. Over 55 percent (6556) of the Anglo high school students in Austin attend one of the 3 schools that is over four-fifths Anglo.
2. Segregative actions taken with segregative intent. It has been the AISD’s policy to assign students to the schools closest to their homes. The City of Austin, with the exception of the strip between East and West Austin, has ethnically segregated housing patterns.12 Hence, the natural, foreseeable, and inevitable result of the AISD’s student assignment policy has been segregated schools throughout most of the city. Moreover, as we found in Austin I, “[affirmative action to the contrary would have resulted in desegregation”. 467 F.2d at 863. The inference is inescapable: the AISD has intended, by its continued use of the neighborhood assignment policy, to maintain segregated schools in East and West Austin.13 The plaintiffs have therefore established a prima facie case of de jure segregation of Mexican-Americans in all portions of the school district except the residentially integrated central city area.14
*391E. The AISD’s Attempted Rebuttal of the Prima Facie Showing of Segregative Intent
The AISD offers numerous arguments to justify the acts that we criticized in Austin I as segregating Mexican-American students in the Austin school system. For the second time, we reject these arguments.
The AISD contends that Mexican-Americans were segregated before 1950 not because of their ethnic background but because they had language difficulties or were the children of migrant workers and needed special educational considerations. We answered this argument in Austin I:
We are not convinced that, to meet the special educational needs of Mexican-American children, the AISD had to keep these children in separate schools, isolate them in Mexican-American neighborhoods, or prevent them from sharing in the educational, social, and psychological benefits of an integrated education.
467 F.2d at 869. We concluded that the AISD intentionally acted to segregate Mexican-Americans in the pre-Brown years.15
The AISD then argues that, even if the early special programs are viewed as intentional segregation, no causal relationship exists between them and the present Mexican-American concentrations in the schools. We rejected this argument in Austin I when we held that the post-1950 AISD actions perpetuated the pre-1950 segregation.16 We now reaffirm our previous rejection of this AISD contention.
The AISD’s primary argument with regard to its post-1950 actions is that, although the location of new schools and the drawing of attendance zones for those schools had “the inevitable and unavoidable result” of increasing the concentrations of Mexican-Americans in the East Austin schools, this segregation resulted from the preexisting residential patterns and not from segregative motives of the AISD. This Court recently rejected a similar argument in Morales v. Shannon, 5 Cir. 1975, 516 F.2d 411, 413, cert. denied, 1975, 423 U.S. 1034, 96 S.Ct. 566, 46 L.Ed.2d 408:
the imposition of the neighborhood assignment system froze the Mexican-American students into the Robb and An-thon schools. There could have been no other result and this is strong evidence of segregatory intent.
See also United States v. Midland Independent School District, 5 Cir. 1975, 519 F.2d 60, cert. denied, 1976, - U.S. -, 96 S.Ct. 1106, 47 L.Ed.2d 314; United States v. Jefferson County Board of Education, 5 Cir. 1966, 372 F.2d 836, 876, 879-80, aff’d en banc, 1967, 380 F.2d 385, cert. denied, 1967, 389 U.S. 840, 88 S.Ct. 67, 19 L.Ed.2d 103.
In rejecting for a second time these contentions of the AISD, we reaffirm—hopefully for the last time—the words of United States v. Midland Independent School District, 519 F.2d at 64:
the facts in the Austin and Corpus Christi cases, however, as in this case, show an overriding intent by the school boards in those districts to isolate, to segregate, Mexican-Americans and blacks.
Finally, we think it important to draw attention to a basic misconception of the AISD, on which a great deal of its argument relies. This misconception goes to the heart of the responsibilities of school authorities to provide equal educational opportunities for the students in their districts. The AISD has argued that “[u]nder Keyes, the school district was prohibited *392from segregating Mexican-American students, but it was under no duty to take affirmative action to attempt to avoid Mexican-American concentrations in the schools which resulted from residential concentrations”. At least in the Texas schools, where we have held that Mexican-American students are entitled to the same benefits of Brown as are blacks, school authorities may not constitutionally use a neighborhood assignment policy that creates segregated schools in a district with ethnically segregated residential patterns. A segregated school system is the foreseeable and inevitable result of such an assignment policy. When this policy is used, we may infer that the school authorities have acted with seg-regative intent.
The segregation is de jure and unconstitutional because it is the result of school board action taken with the obvious (though not necessarily predominant) intent to create or maintain segregated schools. School authorities are then “clearly charged with the affirmative duty to take whatever steps might be necessary to convert to a unitary system in which racial discrimination would be eliminated root and branch”. Green v. County School Board of New Kent County, Virginia, 1968, 391 U.S. 430, 437—38, 88 S.Ct. 1689, 1694, 20 L.Ed.2d 716, 723. As articulated in Austin I, the case before us presents not only the use of a neighborhood assignment policy in a residentially segregated school district, but also the taking of an extensive series of actions dating back to the early twentieth century that had the natural, foreseeable, and avoidable result of creating and maintaining an ethnically segregated school system. The AISD must convert this “still-functioning dual system to a unitary, non-[ethnic] system— lock, stock, and barrel”. United States v. Jefferson County Board of Education, 372 F.2d at 878.
III.
SEGREGATION OF BLACKS
The district court held that “the AISD . has engaged in discriminatory assignment of black students to promote segregation” and ordered the AISD to dismantle its dual school system and convert to an integrated, unitary school system. These holdings have not been challenged on appeal. They are affirmed.
IV.
THE REMEDY
A. The “Desegregation Plan” Adopted by the District Court
1. The Plan. The district court adopted whole the Sixth Grade Center Plan submitted by the AISD. We begin our analysis of this plan by stating what the AISD did not attempt to accomplish through it. The AISD views the junior and senior high schools in Austin as totally desegregated and, therefore, its plan does not further integrate those schools. The AISD, as noted above, does not believe that it has the duty to desegregate the Mexican-Americans and, hence, its plan has only an incidental effect on these students. Finally, the AISD contends that complete desegregation of the elementary schools would require “massive crosstown busing” of 6-10 year olds, which it views as undesirable, and, therefore, its desegregation plan is limited to the sixth grade.
As the AISD describes it,
[t]he Sixth Grade Center Plan essentially establishes six elementary schools in different geographic parts of the School District as centers for all sixth-grade students in the School District. Those buildings which are not serving as elementary schools and would become the sixth-grade centers would be emptied of all students K through grade 5 so the building would be available for the Sixth Grade Center. Students in those schools would be assigned to the nearest available elementary school.
The plan would also set up sixth grades at two of the junior high schools in Austin. Of the six Sixth Grade Centers, two would have Anglo populations of over 80 percent; the sixth grade populations at the two junior high schools would be about 97 percent minority. The AISD estimates that the *393plan would require the busing of about 1900 students, and that about 62 percent of those students would be Anglo.
In an effort to provide equal educational opportunities for all of its students, the AISD has also approved the employment of two assistant superintendents, one to be Mexican-American and one to be black; established majority-to-minority transfer provisions for both black and Mexican-American students; begun to develop a bilingual educational program; made several changes in boundary lines assertedly to produce a better racial and ethnic composition in the city schools; and established an advisory committee to investigate and propose programs for minority students that may be used in Austin.
2. The Plan’s deficiences. As we did in Austin I, we congratulate the AISD for some of the creative educational techniques it has proposed and adopted for equalizing educational opportunities of minority students in Austin. We cannot applaud, however, the channeling of the AISD’s creative abilities into new methods of circumventing its “affirmative duty to take whatever steps might be necessary to convert to a unitary system in which racial [and ethnic] discrimination would be eliminated root and branch”. Green v. County School Board of New Kent County, Virginia, 1968, 391 U.S. 430, 437-38, 88 S.Ct. 1689, 1694, 20 L.Ed.2d 716, 723. The first elementary school “desegregation plan” that the AISD presented to the district court provided for meetings of students one week per month to participate in certain cultural activities. We reversed the district court’s adoption of this plan, holding that “[p]art-time desegregation does not meet constitutional requirements”.17 467 F.2d at 872. On remand, the district court adopted the AISD’s new “desegregation plan”, which leaves untouched the students in grades K-5 and 7-12. For reasons similar to those that underlay our rejection of the AISD’s plan in Austin I, we again hold that the AISD-district court plan is constitutionally deficient. The constitutional duty of the school authorities is to establish a unitary system, not a unitary grade.
The AISD offers two arguments in support of its failure to desegregate grades K to 5. Both are meritless.
First it cites the Supreme Court’s holding that
the district judge or school authorities should make every effort to achieve the greatest possible degree of actual desegregation, taking into account the practicalities of the situation.
Davis v. Board of School Commissioners of Mobile County, 1971, 402 U.S. 33, 37, 91 S.Ct. 1289, 1292, 28 L.Ed.2d 577, 581. The only “practicality” it specifies is the vague, conclusory, and unsupported assertion that children under 10 years old should not be bused for the purpose of desegregation. But busing, a “normal and accepted tool of educational policy”, cannot be rejected without an evidentiary showing that “the time or distance of travel is so great as to risk either the health of the children or significantly impinge on the educational process”. Swann v. Charlotte-Mecklenburg Board of Education, 1971, 402 U.S. 1, 29, 30-31, 91 S.Ct. 1267, 1283, 28 L.Ed.2d 554, 575.
The AISD’s only other defense of the exclusion of kindergarten-to-fifth-grade *394students from its desegregation plan is that the black intervenors and the United States should be precluded from objecting to the Sixth Grade Center Plan because they submitted no plan of their own. The Mexican-American intervenors, however, did propose a desegregation plan, in which the black intervenors concurred. And, as to the United States, although we are disappointed by their noncompliance with the district judge’s request that they submit a plan, we find no basis for denying them the right to criticize the plan submitted by the AISD.
The plan adopted by the district court also fails to comply with the mandate of Austin I. The eight concurring judges in that case held that
[wjhere a student assignment plan is found to be unconstitutional, as here, because of the existence of segregation which has been imposed by statute or by official act against blacks and an identifiable ethnic group (here the Mexican-American students), it is the duty of the school officials to forthwith formulate and implement such student assignment plan as will remedy the discrimination which has been found to exist.
467 F.2d at 884. We held in Parts II and III of this opinion that official discrimination against blacks and Mexican-Americans has infected almost the entire Austin school system. The discrimination has prevented most minority students in the district from securing educational opportunities equal to those of their Anglo counterparts. The AISD’s submission of a “desegregation plan” that would provide an integrated education for only sixth grade students simply does not fulfill the AISD’s duty to remedy that discrimination.
The plan submitted by the AISD would assign students in grades K to 5 to the schools closest to their homes. The district court’s adoption of this plan is directly contrary to the holding in Austin I that
[i]t is apparent that [assignment on a strict neighborhood basis] will not suffice in the AISD although it may suffice as to some schools. To the extent that it does not suffice, the district court will proceed to employ other methods of desegregation.
The Austin I majority also held that if, after trying the pairing or clustering of schools, the realignment of school assignment zones, and the relocation of portable school rooms, “proscribed segregated schools still exist, the court must consider the pairing or clustering of schools in non-contiguous school zones”. 467 F.2d at 885. It was an abuse of discretion for the court to refuse to give serious consideration to the last desegregation method despite the concession of the AISD that
[c]ountless efforts by school officials, consultants, and visiting teams have found it impossible to produce significant desegregation by boundary line changes, contiguous pairing of schools, magnet schools, or other effective means short of massive crosstown busing incident to non-contiguous pairing of ... schools 18
3. The closing of Anderson High School and Kealing Junior High School. In his first opinion in this case, the district judge ordered the closing of two all-black schools, Anderson and Kealing.19 The stu*395dents from those schools were to be transferred to other schools in the system. Six judges concluded in Austin I that the schools were closed for racial reasons and, hence, the closings were unacceptable. 467 F.2d at 872. The remaining eight judges did not consider this issue. On remand, the district court found that the school closings were based on nonraeial considerations. This finding is clearly erroneous. The AISD concedes, as it must, that a primary reason for the school closings was the fear that whites would flee the school system rather than send their children to these East Austin schools. It is hardly a new principle of constitutional law that this fear is an impermissible basis for closing public schools. See, e. g., United States v. Hendry County School District, 5 Cir. 1974, 504 F.2d 550, 553.
Kealing Junior High School must therefore be reopened and used as part of the regular public school program of the District. The district court approved the conversion of Anderson High School into Austin Community College, and the conversion has already taken place. Because it has closed Anderson as a high school, the AISD on remand should present a program that will permit the burdens of desegregation to be as fairly distributed as they would have been if Anderson had not been converted into a community college.20
B. The Finger Plan
1. The Plan. The Mexican-American in-tervenors submitted a desegregation plan prepared by Dr. John A. Finger, Jr., a professor of education at Rhode Island College.21 The “Finger Plan" would convert the school system to a 4-4-4 grade structure, that is, elementary schools would contain grades K to 4, middle schools would contain grades 5 to 8, and high schools would continue to operate grades 9 to 12. All students in grades K to 4 in elementary schools that are over 50 percent minority would be bused to elementary schools that are over 90 percent Anglo. Fifth-to-eighth-grade students in schools that are over 90 percent Anglo would be bused to schools that are over 50 percent minority. The practical effect of the Plan is that kindergarten-to-fourth-grade students in East Austin would be bused to West Austin and fifth-to-eighth-grade students in West Austin would be bused to East Austin. Elementary and junior high schools that are between 50 and 90 percent Anglo are defined as “naturally desegregated” and would remain unchanged. When changing demographic patterns cause any of these schools to fall outside of the “naturally desegregated” range, the schools would be brought within the Finger Plan 4r-4-4 system. The high schools would be integrated by selecting, for each high school, feeder schools that would maximize the integration of that high school. Dr. Finger estimates that 18,659 (the AISD says 25,000) of Austin’s public school students would be bused under his plan.22
2. The AISD’s objections to the Finger Plan. The AISD’s first objection to *396the Finger Plan is that it is counterproductive in that it requires kindergarten-to-fourth-grade Anglo students attending schools in minority areas to be bused along with minority students to schools in Anglo areas; it also requires fifth-to-eighth-grade minority students going to predominantly Anglo schools to be bused along with their Anglo classmates to minority areas. These results are dictated by the feature of the Finger Plan that requires all students in the relevant grades at “sending schools” to be bused to the designated “receiving schools”. About 357 Anglos presently attending minority schools and about 168 minority students presently attending Anglo schools would be bused to the new schools.23 These students represent only about 1 percent of the pre-high school students in Austin. These percentages are simply not substantial enough to invalidate the entire desegregation plan. If, on remand, the district court concludes that a constitutionally sufficient degree of desegregation can be achieved without busing these 525 students across town, the Finger Plan may be so modified.
The AISD also criticizes the Finger Plan because the newly created elementary and middle schools would be (by the AISD’s estimate) about 54 percent minority, although the entire Austin pre-high school system is only about 40 percent minority. This discrepancy is due to the fact that the “naturally desegregated” schools left untouched by the Finger Plan are substantially more than 60 percent Anglo (the system-wide ratio). Thus, Dr. Finger would permit a disproportionately large number of Anglo students to remain at their present schools.
The Supreme Court has held that “[t]he constitutional command to desegregate schools does not mean that every school in every community must always reflect the racial composition of the school system as a whole”. Swann, 402 U.S. at 24, 91 S.Ct. at 1280, 28 L.Ed.2d at 571. But the Court later held in the same opinion that “[t]he district judge or school authorities should make every effort to achieve the greatest possible degree of actual desegregation . . . .” 402 U.S. at 26, 91 S.Ct. at 1281, 28 L.Ed.2d at 572. Such an effort must be made by the district court on remand.24
The AISD also argues that the 4-4-4 school system, though perhaps logical for the purpose of facilitating school desegregation, is basically inconsistent with sound educational principles. This argument is based solely on the testimony of Dr. Jack Davidson, the Superintendent of Schools for the AISD, that placing fifth graders in the same schools (the middle schools) with students four years older “at that period of time—it is the developmental age—produces all kinds of problems”. Even if that statement is considered persuasive, these “problems” can be solved when a final plan is constructed on remand. Dr. Finger testified that his plan could, and perhaps should, be modified to a 5-3-4 system. This plan would replace the middle schools with junior high schools housing the sixth, seventh, and eighth grades.
The AISD next brings to our attention several problems that would be created by the Finger Plan busing program. It *397first argues that the Plan would require the busing of students “in a basic east-west pattern through a traffic system which provides no adequate east-west arteries”. Moreover, the AISD continues, the students would have to be bused through the large complex of the downtown business area, the state office buildings, and the University of Texas, and this would produce a highly congested traffic situation. The AISD also cites the economic cost of the busing, the difficulty of obtaining sufficient fuel, and the inevitability of “white flight”, which would render the Plan ineffective as a desegregation device.25
We think it is important to point out first the reason these remedial costs are relevant to judicial decisionmaking in a school desegregation case. We point this out because the AISD seems to be arguing that these costs are relevant to the determination whether there is a constitutional violation, that is, that the court must decide that the harmfulness of the school segregation is sufficient to justify the remedial costs of correcting that segregation. See generally Fiss, The Jurisprudence of Busing, 39 Law & Contemp.Prob. 194 (1975). We disagree.
The Supreme Court stated the controlling principle in Swann, 402 U.S. at 15-16, 91 S.Ct. at 1276, 28 L.Ed.2d at 566:
a school desegregation case does not differ fundamentally from other cases involving the framing of equitable remedies to repair the denial of a constitutional right. The task is to correct, by a balancing of the individual and collective interests, the condition that offends the Constitution.
In other words, there are two separate phases to a school desegregation case. First, the Court must determine whether there is de jure segregation. This decision, in cases such as the one before us, conforms with the standards of Keyes. Second, the Court must decide upon a remedy. It is at this point that the balancing of interests becomes relevant.26 In this phase of the *398case, the Court must determine the least costly method of correcting the constitutional violation.27 But the above quote from Swann leaves no doubt that, however the balancing of interests is resolved, the constitutional violation must be corrected.
We therefore direct the district court, in completing the desegregation plan for Austin, to minimize the economic cost of busing, the traffic congestion that the busing plan will cause, the time that school children must spend on the buses, and the number of students who will leave the public school system rather than participate in the desegregation plan.28 These costs should be minimized, however, only in the context of a plan that would achieve a constitutionally sufficient degree of desegregation in the Austin school system. The overriding judicial goal must be “the development of a decree ‘that promises realistically to work, and promises realistically to work now.’ ” Davis v. Board of School Commissioners of Mobile County, 402 U.S. at 38, 91 S.Ct. at 1292, 28 L.Ed.2d at 581, quoting Green v. County School Board of New Kent County, Virginia, 391 U.S. at 439, 88 S.Ct. at 1694, 20 L.Ed.2d at 724.
C. Formulation of the Desegregation Decree
We affirm the district court order that the AISD continue in its active efforts to recruit Mexican-American teachers. The AISD should work “toward the achievement, as a goal, of a ratio of mexican-amer-ican teachers to total faculty that approaches the ratio of mexican-american students to the total student population”. Cisneros, 467 F.2d at 151-52. Moreover, the ratio of Mexican-American to Anglo teachers in each school should be substantially the same as it is throughout the district. See United States v. Montgomery County Board of Education, 1969, 395 U.S. 225, 89 S.Ct. 1670, 23 L.Ed.2d 263. We have already held that the AISD has adequately desegregated its faculty on a black-white basis. Austin I, 467 F.2d at 870 n. 37.
The AISD has an ongoing bilingual-bicultural education program that the Superintendent of Schools testified would continue “regardless of the level of federal funding”. Indeed, state and federal law require as much. See 20 U.S.C. § 1703(f); Tex.Educ.Code Ann. § 21.451 et seq. (1975 pocket part). See also Lau v. Nichols, 1974, 414 U.S. 563, 94 S.Ct. 786, 39 L.Ed.2d 1. The district court properly made this commitment a part of its decree.
We held in United States v. Board of Public Instruction of Polk County, Florida, 5 Cir. 1968, 395 F.2d 66, 69, that
[t]here is an affirmative duty, overriding all other considerations with respect to the locating of new schools, except where inconsistent with “proper operation of the school system as a whole” to seek means to eradicate the vestiges of the dual system.
See also Swann, 402 U.S. at 20-21, 91 S.Ct. at 1278, 28 L.Ed.2d at 569; Tasby v. Estes, 5 Cir. 1975, 517 F.2d 92, 104-06, cert. denied, 1975, 423 U.S. 939, 96 S.Ct. 299, 46 L.Ed.2d 271. The district court was therefore correct in incorporating into its order the commitment of the AISD to locate newly constructed schools in such a manner as to maximize integration. When formulating the Austin desegregation decree on remand, the district court should approve new school sites only if they would operate, within the context of the new desegregation decree, to maximize integration in the district.
*399We suggest that the district court consider appointing a master to draft a comprehensive tri-ethnic desegregation plan consistent with this opinion and the decisions of the United States Supreme Court.29 The plan should conform to one of the approaches outlined by Dr. Finger in his written submission of August 14, 1972, and in his testimony.
V.
CONCLUSION
Finally, the intervenors are entitled to reasonable attorneys’ fees. See § 718 of Title VII of the Emergency School Aid Act of 1972, 20 U.S.C. § 1617; Bradley v. School Board of Richmond, 1974, 416 U.S. 696, 94 S.Ct. 2006, 40 L.Ed.2d 476; Henry v. Clarksdale Municipal Separate School District, 5 Cir. 1973, 480 F.2d 583. The district court should conduct evidentiary proceedings to determine the proper amount of fees to be awarded.
We have today held, for the second time, that a desegregation plan submitted by the AISD is constitutionally insufficient. Blacks and Mexican-Americans in Austin have waited a long time for the unitary school system that the constitution requires. We suggest that the district court move expeditiously on remand to provide Austin minority students with such a system.
We reverse the judgment of the district court and remand the case for further proceedings consistent with this opinion. The mandate of the Court shall issue forthwith. The district court should consider appointing a master to prepare a comprehensive desegregation plan. The desegregation plan adopted by the district court in 1973 may be continued only as a stop-gap.
AFFIRMED IN PART, REVERSED IN PART, AND REMANDED.
APPENDIX
AUSTIN INDEPENDENT SCHOOL DISTRICT ETHNIC COMPOSITION OF STUDENTS* 1975-76
School Mexican-Total Black American Anglo
SENIOR HIGH SCHOOLS
2156 (89) Anderson 2432 ro VO H w VO
210 502 (12) (27) 1130 (61) Austin 1842
239 299 (8) (10) 2557 (82) Crockett 3095
388 127 (23) (8) 1141 (69) L. B. Johnson 1656
423 1003 (29) (70) 15 (1) Johnston 1441
291 151 (13) (6) 1843 (81) Lanier 2285
94 203 (7) (14) 1110 (79) McCallum 1407
*400School Total Mexican-Black American Anglo
SENIOR HIGH SCHOOLS
Reagan 1688 502 (30) 171 (10) 1015 (60)
Travis 1900 154 (8) 803 (42) 943 (50)
SENIOR HIGH SCHOOLS TOTALS 17,746 2520 (14) 3316 (19) 11,910 (67) JUNIOR HIGH SCHOOLS
Allan 863 268 (31) 575 (67) 20 (2,)
Bedichek 1168 75 (6) 181 (16) 912 (78)
Burnet 1029 139 (13) 70 (7) 820 (80)
Dobie 1110 206 (18) 118 (11) 786 (71)
Fulmore 921 83 (9) 431 (47) 407 (44)
Lamar 750 50 (7) 133 (18) 567 (75)
Martin 957 74 (8) 851 (89) 32 (3)
Murchison 873 109 (13) 18 (2) 746 (85)
O. Henry 694 57 (8) 80 (12) 557 (80)
Pearce 1308 435 (33) 104 (8) 769 (59)
Porter 934 105 (11) 147 (16) 682 (73)
JUNIOR HIGH SCHOOLS TOTALS 10,607 1601 (15) 2708 (26) 6298 (59)
ELEMENTARY SCHOOLS
Allison 714 106 580 28 (15) (81) (4)
Andrews 515 129 36 350 (25) (7) (68)
Baker 489 47 102 340 (10) (21) (69)
Barrington 654 13 80 561 (2) (12) (86)
317 4 11 302 (1) (4) (95) Barton Hills
*401School Total Mexican-Black American Anglo
ELEMENTARY SCHOOLS
122 (17) 521 (73) Becker 711 00 o VO I — c
436 (97) Blackshear 450
346 (56) 207 (34) Blanton 616
426 (78) 22 (4) Brentwood 547 H VO 00 vo
3 (1) 377 (97) Brooke 389
119 (22) 151 (27) Brown 548 00 H r- in
20 222 (8) (91) 2 (1) Bryker Woods 244
460 (98) Campbell 468 CN h w
20 (3) Casis 591 r-t-co tr> m
557 (83) 26 (4) 85 (13) Cook 668
706 (88) 10 (1) 90 (11) Cunningham 806
210 (30) 41 (6) 442 (64) Dawson 693
104 (91) 3 (3) 7 (6) Dill 114
609 (97) 10 (2) 7 (1) Doss 626
30 (4) 167 (24) 501 (72) Govalle 698
360 (90) 11 (3) 29 (7) Graham 400
474 (99) 7 (1) Gullett 481
398 (68) 120 (20) 72 (12) Harris 590
391 (96) 4 (1) Highland Park 408 CO CO
540 (98) 5 (1) Hill 551 VO rl
723 (76) 94 (10) 130 (14) Joslin 947
177 (68) 47 (18) Lee 260
*402Mexican-School Total Black American Anglo
ELEMENTARY SCHOOLS
478 (77) 47 (7) 99 (16) Linder 624
33 (10) 280 (79) Maplewood 353 O rH H
157 (61) 31 (12) 69 (27) Mathews 257
327 (86) 12 (3) Menchaca 381 H I-* to
4 (1) 4 (1) 487 (98) Metz 495
309 (98) 7 (2) Norman 316
561 (96) 3 (1) 18 (3) Oak Hill 582
303 (94) 18 (6) Oak Springs 321
767 (74) 25 (3) 240 (23) Odom 1032
243 (61) 145 (36) Ortega 399 U> H
394 (94) Palm 419 tO
154 (61) 47 (18) Pease 254 ro r-H in cn
279 (52) 205 (38) 53 (10) Pecan Springs 537
500 (93) 30 (6) Pillow 538
627 (76) 172 (21) Pleasant Hill 821
574 (84) 37 (6) Read 681 O O P- r-H
196 (71) 4 (1) 78 (28) Reilly 278
115 (50) 2 (1) 111 (49) Ridgetop 228
207 (78) 7 (3) 52 (19) Rosedale 266
156 (98) Rosewood 160
622 (64) 32 (3) St. Elmo 975 w to U> H
5 (1) 455 (92) 494 Sims
*403Mexican-School Total Black American Anglo
ELEMENTARY SCHOOLS
Summitt 201 200 (100)
Sunset Valley 611 20 (3) 61 (10) 530 (87)
Travis Heights 762 93 (12) 301 (40) 368 (48)
Walnut■Creek 308 . 3 (1) 42 (14) 263 (85)
Webb 897 169 (19) 122 (14) 606 (67)
Winn 544 180 (33) 34 (6) 330 (61)
Woolridge 747 32 (4) 80 (11) 635 (85)
Wooten 635 18 (3) 85 (13) 532 (84)
Zavala 415 23 (6) 384 (92) 8 (2)
Zilker 541 23 (4) 100 (19) 418 (77)
ELEMENTARY SCHOOLS TOTALS 30,567 4989 (16) 7242 (24) 18,336 (60)
GRAND TOTAL 58,920 9110 (15) 13,266 (23) 36,544 (62)
. The term “minority” is used to refer collectively to Mexican-American and black students.
. Keyes v. School District No. 1, Denver, Colorado, 1973, 413 U.S. 189, 93 S.Ct. 2686, 37 L.Ed.2d 548.
. The eight concurring judges in Austin I held:
The power of the district court will depend first upon a finding of the proscribed discrimination in the school system. ... In determining the fact of discrimination vel non . . . , the district court must identify the school or schools which are segregated as a result of such discrimination. . . . The importance of such a determination will be seen in some populous school districts embracing large geographical areas. There may be segregated schools which are the result of unconstitutional statutes or of official action. There may be other one race schools which are the product of neutral, non-discriminatory forces.
467 F.2d at 884. To the extent that this holding requires a court to identify the intentional state
action that segregated a school as a prerequisite to including that school in a desegregation plan, the holding was unambiguously supervened by Keyes. The Supreme Court there stated:
We have never suggested that plaintiffs in school desegregation cases must bear the burden of proving the elements of de jure segregation as to each and every school or each and every student within the school system. . . . [Wjhere plaintiffs prove that the school authorities have carried out a systematic program of segregation affecting a substantial portion of the students, schools, teachers, and facilities within the school system, it is only common sense to conclude *387that there exists a predicate for a finding of the existence of a dual school system. 413 U.S. at 200, 201, 93 S.Ct. at 2694, 37 L.Ed.2d at 558.
. Prosser states the tort rule that “[i]ntent extendfs] not only to those consequences which are desired, but also to those which the actor believes are substantially certain to follow from what he does”. The Law of Torts § 8, at 31 (4th ed. 1971). The rule has also been applied in many other areas. See, e. g., NLRB v. Great Dane Trailers, 1967, 388 U.S. 26, 33, 87 S.Ct. 1792, 1797, 18 L.Ed.2d 1027, 1034 quoting NLRB v. Erie Resistor Corp., 1963, 373 U.S. 221, 227, 228, 231, 83 S.Ct. 1139, 1144, 1145, 1147, 10 L.Ed.2d 308, 313, 314, 316 (discrimination against labor union member in violation of § 8(a)(3) of the National Labor Relations Act):
Some conduct ... is so‘inherently destructive of employee interests’ that it may be deemed proscribed without need for proof of an underlying improper motive. . That is, some conduct carries with it ‘unavoidable consequences which the employer not only foresaw but which he must have intended’ and thus bears ‘its own indicia of intent.’
Gomillion v. Lightfoot, 1960, 364 U.S. 339, 341, 347, 81 S.Ct. 125, 127, 130, 5 L.Ed.2d 110, 113, 116 (unconstitutionality of state statute redefining the boundaries of the City of Tuskegee). There, based on its observation that the “inevitable effect” of the redefinition of the City’s boundaries was to remove almost all of its black voters, the Court observed that “the legislation is solely concerned with” segregating whites and blacks so as to deprive blacks of their vote, and that, “to that end [the Legislature] has incidentally changed the city’s boundaries”. (emphasis added). Miller v. Milwaukee, 1927, 272 U.S. 713, 715, 47 S.Ct. 280, 71 L.Ed. 487, 489 (validity of indirect state tax on federally tax-exempt income): “A result intelligently foreseen and offering the most obvious motive for an act that will bring it about, fairly may be taken to have been a purpose of the act.”
. See Brest, 1971 Sup.Ct.Rev. at 120-21.
The process does not differ from that of inferring ultimate facts from basic facts in other areas of the law. It is grounded in an experiential, intuitive assessment of the likelihood that the decision was designed to further one or another objective.
Id. at 121. See also Comment, Sex Separation in School Desegregation Plans, 37 V.Chi.L.Rev. 296, 301-05 (1970); Developments in the Law—Equal Protection, 82 Harv.L.Rev. 1065, *3891077 (1969). Indeed, in Keyes the Supreme Court inferred the School Board’s segregative intent with respect to one section of Denver (the core city) from evidence of intentional segregation in another area (Park Hill).
. We are not the first circuit to read the “natural and foreseeable consequences” test into the Keyes requirement of segregative intent. See, e. g., Hart v. Community School Board of Education, New York School District # 21, 2 Cir. 1975, 512 F.2d 37, 50-51; Morgan v. Kerrigan, 1 Cir. 1974, 509 F.2d 580, cert. denied, 1975, 421 U.S. 963, 95 S.Ct. 1950, 44 L.Ed.2d 449; Oliver v. Michigan State Board of Education, 6 Cir. 1974, 508 F.2d 178, 182, cert. denied, 1975, 421 U.S. 963, 95 S.Ct. 1950, 44 L.Ed.2d 449. But see Soria v. Oxnard School Board of Trustees, 9 Cir. 1973, 488 F.2d 579, 585, cert. denied, 1974, 416 U.S. 951, 94 S.Ct. 1961, 40 L.Ed.2d 301. See generally Fiss, School Desegregation: The Uncertain Path of the Law, 4 Phil. & Pub. Affairs 3 (1974).
. See, e. g., Hart, 512 F.2d at 50; Morgan, 509 F.2d at 585-86; Oliver, 508 F.2d at 187; United States v. Board of School Commissioners of Indianapolis, Indiana, 7 Cir. 1973, 474 F.2d 81, 89, cert. denied, 1973, 413 U.S. 920, 93 S.Ct. 3066, 37 L.Ed.2d 1041.
. Justice Powell, in a separate opinion, made the following observations about the approach of the Keyes majority: The Court “searches for de jure action in what the Denver School Board has done or failed to do”. Keyes, 413 U.S. at 230, 93 S.Ct. at 2708, 37 L.Ed.2d at 575. “Every act of a school board and school administration, and indeed every failure to act where affirmative action is indicated, must now be subject to scrutiny.” 413 U.S. at 234, 93 S.Ct. at 2710, 37 L.Ed.2d at 578.
. Arvizu v. Waco Independent School District, 5 Cir. 1974, 495 F.2d 499, 505 n. 10. See also Keyes, 413 U.S. at 197, 93 S.Ct. at 2691, 37 L.Ed.2d at 556.
. In citing these statistics, we recall the words of United States v. Jefferson County Bd. of Educ., 5 Cir. 1966, 372 F.2d 836, 887, aff’d en banc, 1967, 380 F.2d 385, cert. denied, 1967, 389 U.S. 840, 88 S.Ct. 67, 19 L.Ed.2d 103:
A similar inference [of deliberate discrimination against Negroes] may be drawn in school desegregation cases, when the number of Negroes attending school with white children is manifestly out of line with the ratio of Negro school children to white school children in public schools.
. Another 18 percent (4358) of the Anglo elementary and junior high school students attend one of the 8 schools that is between 75 and 80 percent Anglo.
. East Austin is bordered on the north by East Nineteenth Street and the airport, on the south by the Colorado River, on the west by Interstate Highway 35, and on the east by the AISD boundary line. We found in Austin I that 64 percent of the City’s Mexican-Americans live in East Austin. A large portion of the remaining Mexican-Americans live in the area between East and West Austin.
. The same conclusion is inferable from other evidence as well. We held in Austin I
that the AISD has, in its choice of school site locations, construction and renovation of schools, drawing of attendance zones, student assignment and transfer policies, and faculty and staff assignments, caused and perpetuated the segregation of Mexican-
American students within the school system. 467 F.2d at 865-66. We also found that “[t]he natural and foreseeable consequence of these actions was segregation of Mexican-Americans”. 467 F.2d at 863. The Supreme Court inferred segregative intent from the same kind of circumstantial evidence in Keyes. See 413 U.S. at 192, 93 S.Ct. at 2689, 37 L.Ed.2d at 553. The inference of segregative intent that the Supreme Court made regarding the Denver school authorities is equally applicable to their counterparts in Austin.
. The district court held that the AISD had, in the past, assigned black students for the purpose of promoting segregation. The plaintiffs argue that this finding of de jure segregation in a substantial portion of the Austin school district triggers the Keyes presumption of unlawful segregation in the remainder of the district. The AISD responds that this Keyes presumption is inapplicable to the case before us because “[t]he existence of a statutorily based black-white system has no probative value with respect to concentrations of Mexican-American students when the Mexican-American students were classified and treated as white under the dual system”. See Higgins v. Bd. of Educ. of *391Grand Rapids, 6 Cir. 1974, 508 F.2d 779, 789. We need not resolve this dispute about the Keyes presumption because, even without this presumption, we conclude that the AISD has taken actions intentionally calculated to segregate the Mexican-American students throughout the district.
. [TJhe AISD used dual-overlapping attendance zones, student assignment policies, and site selection to segregate Mexican-American students in the years prior to 1954.
467 F.2d at 867.
. After the Supreme Court decision in Brown, the AISD nominally undertook to abolish the dual system based on separate schools for blacks and whites. But the board continued to perpetuate segregation of Mexican Americans.
467 F.2d at 867.
. The federal courts may adopt desegregation remedies requiring busing only as a last resort. See 20 U.S.C. §§ 1713, 1755. In the case before us, however, we find that crosstown busing is the only desegregation method that will work. This finding is supported not only by the above-quoted statement of the AISD but also by the residential patterns in Austin. In school districts with segregated neighborhoods, “[d]esegregation plans cannot be limited to the walk-in school”. Swann v. Charlotte-Mecklenburg Bd. ofEduc., 1971, 402 U.S. 1, 30, 91 S.Ct. 1267, 1283, 28 L.Ed.2d 554, 575. Hence the federal statutes do not bar the court-ordered transportation of students in Austin.
. At the time Anderson High School was closed, it was 98 percent black and its student body constituted about 44 percent of the black high school population in the Austin school district. Kealing Junior High School was also 98 percent black and its student body constituted 46 percent of the black junior high school students in the district. See Austin I, 467 F.2d at 876-77, Appendix A.
. In Tasby v. Estes, 5 Cir. 1975, 517 F.2d 92, cert. denied, 1975, 423 U.S. 939, 96 S.Ct. 299, 46 L.Ed.2d 271, we held another “part-time” desegregation plan constitutionally deficient. The unique feature of the plan submitted there by the Dallas Independent School District (DISD) and adopted by the district court was the requirement that there be a minimum of one hour a day of contact between the races through two-way oral and visual television communication between two or more schools. We held:
The Supreme Court has made it clear that nothing less than the elimination of predominantly one-race schools is constitutionally required in the disestablishment of a dual school system based upon segregation of the races. For this reason, the district court’s elementary school “television plan” must be rejected as a legitimate technique for the conversion of the DISD from a dual to a unitary educational system.
517 F.2d at 103.
. For example, through the construction of a new high school in East Austin.
. Dr. Finger is a recognized authority in the area of school desegregation and has designed the plans presently being used in several cities. He prepared, for example, the plan for Charlotte, North Carolina, which was approved by the Supreme Court in Swann v. Charlotte-Mecklenburg Bd. of Educ., 1971, 402 U.S. 1, 91 S.Ct. 1267, 28 L.Ed.2d 554.
. The Finger Plan would therefore require the busing of about 32 percent (42 percent according to the AISD’s estimate) of the Austin students. This is comparable to the Charlotte-Mecklenburg, North Carolina school system, which, before the Supreme Court’s 1971 desegregation order, planned to bus 27 percent of its students “without regard to desegregation plans”, and the Mobile County, Alabama school system, which bused 30 percent of its students before the Supreme Court’s 1971 desegregation order. See Swann v. Charlotte-Mecklenburg Bd. of Educ., 1971, 402 U.S. 1, 6, 29 n. 11, 91 S.Ct. 1267, 1271, 28 L.Ed.2d 554, 561; Davis v. Board of School Commissioners of Mobile County, 1971, 402 U.S. 33, 34, 91 S.Ct. 1289, 1290, 28 L.Ed.2d 577, 579. Moreover, the Supreme Court noted in Swann that about 39 percent of this country’s public school children were bused to their schools in 1969-70. 402 U.S. at 29, 91 S.Ct. at 1282, 28 L.Ed.2d at 574.
. The AISD has calculated that the correct figures are 535 Anglos and 336 minority students. These numbers are too high because the AISD has assumed that all students at elementary and junior high schools over 50 percent minority or 90 percent Anglo would be bused to new elementary or middle schools outside of their neighborhoods. The Finger Plan, however, would bus only kindergarten-to-fourth-grade students from the elementary schools in East Austin and fifth-to-eighth-grade students from the elementary and junior high schools in West Austin. Hence, the AISD estimates are about one-third too high for the Anglo students at minority schools in East Austin and about one-half too high for the minority students at Anglo schools in West Austin.
. This may require an adjustment of Dr. Finger’s statistical definition of “naturally desegregated” schools. Quotas may be a starting point for the district court, but are not an ironclad requirement. See Milliken v. Bradley, 1974, 418 U.S. 717, 740-41, 94 S.Ct. 3112, 3125, 41 L.Ed.2d 1069, 1088-89; North Carolina Bd. of Educ. v. Swann, 1971, 402 U.S. 43, 46, 91 S.Ct. 1284, 1286, 28 L.Ed.2d 586, 589.
. The AISD contends that whites will flee the Austin public school system to attend private schools and public schools in surrounding school districts. As a result, the AISD concludes that the plan fails to meet the standard of Davis v. Board of School Commissioners of Mobile County, 1971, 402 U.S. 33, 37, 91 S.Ct. 1289, 1292, 28 L.Ed.2d 577, 581: “The measure of any desegregation plan is its effectiveness.” The district court was presented with two desegregation plans, the AISD Plan, which would desegregate only the sixth grade, and the Finger Plan, which would desegregate the entire school system. It is wholly speculative whether white flight will eventually render the Finger Plan less effective than the AISD Plan in transforming the AISD into a unitary system. It is beyond dispute, however, that the Finger Plan is the more effective desegregation device for the immediate future. For this reason, and others that we have specified in this opinion, it was an abuse of discretion for the district court to adopt the AISD Plan. See United States v. Bd. of School Commissioners of Indianapolis, Indiana, 7 Cir. 1974, 503 F.2d 68, 75-76, cert. denied, 1973, 413 U.S. 920, 93 S.Ct. 3066, 37 L.Ed.2d 1041.
. The Court stated in Brown II that, in determining whether school authorities should be given additional time to carry out the desegregation remedy, “the courts may consider problems related to administration, arising from the physical condition of the school plant, the school transportation system, personnel . . . .” 349 U.S. 294, at 300, 75 S.Ct. 753, 99 L.Ed. 1083. The Brown II Court, however, carefully limited its approval of consideration of these problems to the delay issue. See also Watson v. Memphis, 1963, 373 U.S. 526, 532-33, 83 S.Ct. 1314, 1318, 10 L.Ed.2d 529, 534-35.
The AISD’s arguments that its school district should not be ordered to desegregate “root and branch” because of the economic cost and the specter of white flight have already been rejected by the Supreme Court. In Watson, 373 U.S. at 537-38, 83 S.Ct. at 1320-21, 10 L.Ed.2d 537-38, the Court was unpersuaded by the argument that desegregation of the Memphis parks should be delayed because of the expenses it would generate:
it is obvious that vindication of conceded constitutional rights cannot be made dependent upon any theory that it is less expensive to deny them to afford them. We will not assume that the citizens of Memphis accept the questionable premise implicit in this argument or that either the resources of the city are inadequate, or its government unresponsive, to the needs of all of its citizens.
And the Court has repeatedly held that segre-gative state action must be terminated and remedied despite public disagreement with the constitutional principles. See, e. g., United States v. Scotland Neck City Bd. of Educ., 1972, 407 U.S. 484, 490-91, 92 S.Ct. 2214, 2217-18, 33 L.Ed.2d 75, 80-81; Watson, 373 *398U.S. at 535, 83 S.Ct. at 1319, 10 L.Ed.2d at 536; Cooper v. Aaron, 1958, 358 U.S. 1, 16, 78 S.Ct. 1401, 1408, 3 L.Ed.2d 5, 15; Brown II, 349 U.S. at 300, 75 S.Ct. at 756, 99 L.Ed. at 1106; Buchanan v. Warley, 1917, 245 U.S. 60, 81, 38 S.Ct. 16, 20, 62 L.Ed. 149, 163.
. See Fiss, 39 Law & Contemp.Prob. at 198. Professor Fiss correctly points out that “[t]he court need not choose the remedy that has the best cost-benefit relationship since it may eliminate a smaller portion of the harm”. Id.
. On the issue of “white flight”, the district court should accord appropriate weight to the following testimony of Dr. Finger:
[M]y thought in preparing this plan was to minimize the public anguish over busing as much as possible, that there isn’t any way that one can overcome it, but my attempt was to minimize it as much as possible.
Derived from October 10, 1975 submission of the AISD.Figures in parentheses indicate percentages.
. The AISD should provide staff assistance to the master or expert upon his request. See, e. g., United States v. Bd. of School Commissioners of Indianapolis, Indiana, 7 Cir. 1974, 503 F.2d 68, 78, cert. denied, 1973, 413 U.S. 920, 93 S.Ct. 3066, 37 L.Ed.2d 1041; Bradley v. Milliken, 6 Cir. 1973, 484 F.2d 215, 252, rev’d on other grounds, 1974, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069.