“[OJnce you begin the process of segregation, it has its own inertia. It continues on without enforcement.” 1 This comment by one expert on segregation in schools succinctly summarizes the state of affairs in Topeka. As a former de jure segregated school system, Topeka has long labored under the duty to eliminate the consequences of its prior state-imposed separation of races. Brown v. Board of Educ., 349 U.S. 294, 75 S.Ct. 753, 99 L.Ed. 1083 (1955). The district court concluded that Topeka has fulfilled that duty, and that the school system is now unitary. Because we are convinced that Topeka has not sufficiently countered the effects of both the momentum of its pre-Brown segregation and its subsequent segregative acts in the 1960s, we reverse. Specifically, we hold that the district court erred in placing the burden on plaintiffs to prove intentional discriminatory conduct rather than according plaintiffs the presumption that current disparities are causally related to past intentional conduct. We are convinced that defendants failed to meet their burden of proving that the effects of this past intentional discrimination have been dissipated. We also reverse the district court’s holding that the Topeka school district has not violated Title VI. However, we affirm the court’s dismissal of the Governor of the State of Kansas and its ruling that the State Board of Education bears no liability for segregation in Topeka’s schools.
I.
LEGAL HISTORY
Prior to 1954, a Kansas statute permitted certain cities to maintain separate schools for white and black children below the high school level. In 1941, however, the Kansas Supreme Court held segregation in Topeka’s junior high schools to be unconstitutional. See Graham v. Board of Educ., 153 Kan. 840, 114 P.2d 313 (1941) (separate facilities not equal). Topeka was thus legally permitted to operate segregated schools only at the elementary level. The Topeka Board of Education operated such a system. In 1951, black citizens of Topeka filed a class action challenging the constitutionality of the Kansas law authorizing school segregation. Brown v. Board of Educ., 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954) (Brown I), followed, beginning a new era of American jurisprudence by bringing an end to the doctrine of “separate but equal” and declaring segregation unconstitutional.
The Topeka Board of Education did not wait for the decision in Brown I before taking steps towards desegregating Topeka’s elementary schools. It began that process in 1953 by permitting black students to attend two formerly all-white schools. It then gradually increased the number of schools black students might attend. Accordingly, when the Supreme Court considered the question of the relief *855appropriate in school desegregation eases, it noted that “substantial progress” had already been made in Topeka. Brown v. Board of Educ., 349 U.S. 294, 299, 75 S.Ct. 753, 756, 99 L.Ed. 1083 (1955) (Brown II). On remand, the district court criticized one aspect of the Board’s desegregation plan but described it overall as “a good faith effort to bring about full desegregation in the Topeka Schools in full compliance with the mandate of the Supreme Court.” Brown v. Board of Educ., 139 F.Supp. 468, 470 (D.Kan.1955). The court retained jurisdiction of the case, and the decision was not appealed.
Nineteen years later, in 1974, the Office of Civil Rights (OCR) of the Department of Health, Education, and Welfare (HEW) notified the Topeka school district that it was not in compliance with section 601 of Title VI of the Civil Rights Act of 1964.2 After the Topeka Board of Education failed to adopt a plan designed to remedy the noncomplying conditions identified by OCR, HEW began administrative enforcement proceedings against the Topeka school district. The Board filed suit in federal court and obtained a preliminary injunction against the administrative proceeding on the ground that the district court’s 1955 decision was a final order, and that the school district was still operating under that court order and still subject to the court’s jurisdiction. HEW was thereby precluded from taking administrative action. See generally Brown v. Board of Educ., 84 F.R.D. 383, 390-91 (D.Kan.1979). In 1976, the Board submitted a plan acceptable to HEW, and both the administrative proceeding and the suit in federal court were dismissed. The Board implemented the plan over the next five years.
In 1979, a group of black parents and children sought to intervene in Brown as additional named plaintiffs on the ground that they were members of the original class and that the original named plaintiffs no longer had a sufficient interest in the matter to represent their interests. The intervenors asserted that Topeka has failed to desegregate its schools in compliance with the Supreme Court’s mandate, and that the Topeka school district currently maintains and operates a racially segregated school system. Their request to intervene was granted.3 See Brown, 84 F.R.D. 383. A long discovery and motion stage followed the granting of the intervenors’ motion.
Trial took place in October 1986. The court found the Topeka school district to be an integrated, unitary school system. Brown v. Board of Educ., 671 F.Supp. 1290 (D.Kan.1987). The court also held that the Topeka school district had not violated Title VI of the Civil Rights Act of 1964, dismissed the Governor of Kansas from the case, and found that the State Board of Education bore no liability for racial conditions in the school district. This appeal followed.
II.
BRIEF FACTUAL HISTORY
A. Population Change
In 1950, Topeka’s population was approximately 10% black. While Topeka’s population grew significantly until 1970 and then dropped, the black percentage of the population remained approximately the same. The Hispanic population of Topeka has been slightly less than 5% since 1970. Other minorities make up less than 1.5% of the population.
The distribution of Topeka’s population has changed more significantly than its composition. In general, the outer parts of Topeka, particularly on the western side, have grown considerably in population, while the inner city has declined. Until *856recently, the western side of Topeka was almost exclusively white. The black population of Topeka was concentrated in a few areas in the center of the city in the 1950s; it has since spread widely throughout the eastern part of the city and has gradually begun to move into the western side of Topeka.
The percentage of black and minority children in the Topeka schools has long been higher than the percentage of blacks and minorities in the Topeka population as a whole and has risen over time. In 1952, black students constituted 8.4% of the total number of students in Topeka. By 1966, the percentage of black students in the Topeka school district was 11.6% and the percentage of minority students was 16.0%. In 1975, black students constituted 14.7%, and minority students 20.9%, of the school population. The latest figures used at trial, those for the 1985 school year, showed 18.4% black and 25.95% minority children in the system.
B. Elementary Schools
In 1951, four Topeka elementary schools were reserved for black children, Buchanan, McKinley, Monroe, and Washington. Eighteen elementary schools educated white children. Black children were bused to their schools; white children attended neighborhood schools. 671 F.Supp. at 1291. Under the four-step plan approved by the district court in 1955, all elementary schools were to be opened by September of 1956 to black and white children under a neighborhood school policy. Id. at 1293. As a result of the new neighborhood school policy, three schools remained all- or virtually all-black (Buchanan (100%), Monroe (100%), Washington (99.4%)), and two others became more than 20% black in a school district with a black elementary student population of less than 10%.4 McKinley was closed.
During the late 1950s, the school district acquired by annexation the Avondale (outer Topeka, south) and Highland Park (middle and outer Topeka, east) school districts as well as other territory on the edges of the district. Existing schools within the acquired area were either primarily white or primarily black. As school enrollments grew and the population began to shift, the school district began to close elementary schools in the inner part of the city and open them in the rapidly growing outer part of the city. Two of the closed schools were former de jure black schools (Buchanan, and Washington); another de jure school (McKinley) had already been closed. The new schools were built in the newly acquired white areas and opened with all or virtually all white students.
Racial statistics were not kept in an organized fashion from 1956 to 1966. In 1966, the school district operated thirty-five elementary schools. There were some white students in every school. Minority students were present in thirty-two schools. Nineteen of the schools were 90 + % white. An additional seven schools were 80-90% white. Four schools were more than 50% minority, and a fifth was almost 50%. The highest percentage of minority students was 93.1% (Parkdale), and the lowest was 0% (Lyman, McEa-chron, and Potwin). Sixty-five percent of white students attended 90 + % white schools and an additional 18.7% attended 80-90% white schools. Close to half of all minority students attended 50 + % minority schools.5
A second major reorganization of the elementary schools took place in the late 1970s, after the HEW complaint was filed. Eight elementary schools closed over a six-year period, including the last of the four former de jure black schools (Monroe). In September 1982, when the reorganization had ended, minority and white students were present in each of the district’s twen*857ty-six elementary schools.6 Five schools were 90 + % white, and another seven were 80-90% white. Four schools were 50 + % minority, two of them were schools that had been 50 + % minority since 1966. The highest percentage of minority students was 60.6% (Highland Park North), and the lowest was 3.4% (McClure). Close to one-quarter of all white students attended the 90 + % white schools, and another third attended 80-90% white schools, total-ling 58% overall. The percentage of minority students in 50 + % minority schools was 35.5%.7 This was the status of the elementary schools three years after plaintiffs moved to intervene in this lawsuit on the grounds that the school system had not met its mandate to desegregate.
With one or two exceptions, the relative percentages of white and minority students in the elementary schools have changed only by two or three percentage points since that time. The most significant change is that the schools with the highest white percentages have gained some minority students. Thus, in 1985, the lowest percentage of minority students in any school was 7.2% (McClure).8
C. Secondary Schools
In 1954, the Topeka school district operated six junior highs and one high school. Two schools were 90 + % white, and three were 80 + % white. The estimated percentage of black students at the junior high schools ranged from 1.7% (Roosevelt) to 30% (East Topeka).9 While the dissent emphasizes that segregation was not mandated by law for the secondary schools when Brown I was decided, and the student bodies were racially mixed, Linda Brown Smith testified that the junior high school she attended in 1955 (Curtis) had an all-white faculty.10 When she entered Topeka High school, she recalls that it had one black faculty member.11
During the late 1950s-early 1960s period of annexations and building, two junior high schools joined the school system, and *858three junior highs were built. At the high school level, Highland Park high school was annexed, and Topeka West high school was built. All of these schools were in the newly acquired white outer part of the school district and opened as white or primarily white schools. 671 F.Supp. at 1299.
In 1966, there were thus eleven junior high and three high schools. At that time, the average minority percentage for the junior high and high schools was 15.3% and 14.9%, respectively.12 Of the junior highs, five had 90 + % white students and another three had 80-90% white students; one had 50 + % minority students. The highest percentage of minority students at one school was 61.8% (East Topeka), and the lowest percentage was 0% (Capper). Of the high schools, Topeka High was nearly one-quarter minority, Highland Park High had close to 15% minority students, and Topeka West had .4% minority students. French junior high school opened in 1970 in the southwestern part of the school district as a primarily white school. 671 F.Supp. at 1299.
The reorganization of the late 1970s in response to the HEW complaint included the junior high schools. Two junior highs closed in 1975. In 1980, five more junior highs closed and two schools were opened as the district shifted from a junior high (6-3-3) to a middle school (6-2-4) format. In 1981, after the end of the reorganization, there were six middle schools in the Topeka school district. Two were 90 + % white and one was 80-90% white. The highest percentage of minority students was 45.7% (Eisenhower) and the lowest 5.5% (French). By 1985, the relative percentages at some schools had altered by approximately 5%, but the pattern across the district had not changed.13 The percentage of minority students at the three high schools was 39.8% (Highland Park), 32.5% (Topeka High), and 5.25% (Topeka West) in 1981, and 33.6% (Highland Park), 30.9% (Topeka High), and 7.9% (Topeka West) in 1985.
III.
THE PARTIES
All of the parties to this case have changed. The original plaintiff children have long since left the Topeka school system. The school district has been reorganized, and the State Board of Education came into existence in 1969. These changes have affected the posture of the litigation to some extent. The original named plaintiffs represented black elementary school children and their parents. Current named plaintiffs represent black children throughout the school system and their parents. The school district grew considerably in size as the city of Topeka annexed territory, although the school district’s boundaries were fixed about 1960 while the city continued to grow. The district was also renamed Unified School District # 501 as part of a state-wide reorganization of school districts in 1965. 671 F.Supp. at 1292. The State Board of Education is the product of a 1966 state constitutional amendment. Its powers differ *859considerably from those of its predecessor. Id.) Brief for Individually-Named Defendants Associated with the State Board of Education at 1, 3-4.
IV.
GENERAL PRINCIPLES OF UNITARINESS
Unitariness is a finding of fact reviewed under the clearly erroneous standard.14 Before we assess the status of school desegregation in Topeka, we set forth the principles that guide our consideration of the unitariness issue.
The district court defined a unitary school system as “one in which the characteristics of the 1954 dual system either do not exist or, if they exist, are not the result of past or present intentional segregative conduct of” the school district. 671 F.Supp. at 1293.15 These are necessary ingredients in a unitariness determination because once a violation is found, “[t]he Board has ... an affirmative responsibility to see that pupil [and faculty] assignment policies and school construction and abandonment practices ‘are not used and do not serve to perpetuate or re-establish the dual school system.’ ” Dayton Bd. of Educ. v. Brinkman, 443 U.S. 526, 538, 99 S.Ct. 2971, 2979, 61 L.Ed.2d 720 (1979) (Dayton II) (quoting Columbus Bd. of Educ. v. Penick, 443 U.S. 449, 460, 99 S.Ct. 2941, 2948, 61 L.Ed.2d 666 (1979). An additional essential requirement of unitariness, however, is whether “school authorities [have made] every effort to achieve the greatest possible degree of actual desegregation, taking into account the practicalities of the situation.” Davis v. Bd. of School Comm’rs, 402 U.S. 33, 37, 91 S.Ct. 1289, 1292, 28 L.Ed.2d 577 (1971); see also Swann v. Charlotte-Mecklenburg Bd. of Educ., 402 U.S. 1, 26, 91 S.Ct. 1267, 1281, 28 L.Ed.2d 554 (1971).
To determine whether a school district has become unitary, therefore, a court must consider what the school district has done or not done to fulfill its affirmative duty to desegregate, the current effects of those actions or inactions, and the extent to which further desegregation is feasible.16 After a plaintiff establishes intentional segregation at some point in the past and a current condition of segregation, a defendant then bears the burden of proving that its past acts have eliminated all traces of past intentional segregation to the maximum feasible extent.
A. Current Condition of Segregation
The actual condition of the school district at the time of trial is perhaps the most crucial consideration in a unitariness determination. The plaintiff bears the burden of showing the existence of a current condition of segregation. The case law is decidedly unclear as to the precise meaning of that term.17 In our view, a plaintiff *860must prove the existence of racially identifiable schools, broadly defined, to satisfy the burden of showing a current condition of segregation. Racially identifiable schools may be identifiable by student assignment alone, in the case of highly one-race schools, or by a combination of factors where the school is not highly one-race in student assignment.
“What is or is not a segregated school will necessarily depend on the facts of each particular case. In addition to the racial and ethnic composition of a school’s student body, other factors, such as the racial and ethnic composition of faculty and staff and the community and administration attitudes toward the school, must be taken into consideration.”
Keyes v. School Dist. No. 1, 413 U.S. 189, 196, 93 S.Ct. 2686, 2691, 37 L.Ed.2d 548 (1973).
Although virtual one-race schools “require close scrutiny,” they are not always unconstitutional.18 Swann, 402 U.S. at 26, 91 S.Ct. at 1281. Their existence in a system with a history of de jure segregation, however, establishes a presumption that they exist as the result of discrimination and shifts the burden of proof to the school system. Id. The presence of essentially one-race schools is thus sufficient to satisfy a plaintiffs initial burden of showing a current condition of segregation.
Courts have used various standards to define “one-race schools.”19 Standards may appropriately differ from school district to school district because the percentage of minority students may likewise vary.20 Whatever the minority percentage district-wide, however, it is clear that a school with 90 + % students of one race is a predominantly one-race school.21 Moreover, this is true whether the students at the school in question are white or minority.22
Where racial imbalance in student assignment is still extreme in a system that formerly mandated segregation, appellate courts have reversed findings of unitari*861ness without looking to other factors.23 However, no particular degree of racial balance is required by the Constitution.24 A degree of imbalance is likely to be found in any heterogeneous school system. Therefore, the existence of some racial imbalance in schools will often not be conclusive in itself.
Where numbers alone are insufficient to define racially identifiable schools, courts look to demography, geography, and the individual history of particular schools and areas of the city.25 While a multi-race school cannot be classified as racially identifiable merely by tallying up the race of the students who attend it, such a school may be racially identifiable “simply by reference to the racial composition of teachers and staff, the quality of school buildings and equipment, or the organization of sports activities,” among other factors. Swann, 402 U.S. at 18, 91 S.Ct. at 1277.26 These factors alone can establish a prima facia case of a constitutional violation. Id. Therefore, a plaintiff may prove a school to be racially identifiable by factors that may, but need not, include student assignment.
B. The Parties’ Burdens
Once a plaintiff has proven the existence of a current condition of segregation, the school district bears the substantial burden of showing that that condition is not the result of its prior de jure segregation. Under the relevant Supreme Court decisions, mere absence of invidious intent on the part of the school district is not sufficient to satisfy its “heavy burden” of proof; the district’s duty is to act' affirmatively, not merely to act neutrally.
“[T]he measure of the post-Brown I conduct of a school board under an unsatisfied duty to liquidate a dual system is the effectiveness, not the purpose, of the actions in decreasing or increasing the segregation caused by the dual system. As was clearly established in Keyes and Swann, the Board had to do more than abandon its prior discriminatory purpose. The Board has had an affirmative responsibility to see that pupil assignment policies and school construction and abandonment practices ‘are not used and do not serve to perpetuate or re-establish the dual school system.’ ”
Dayton II, 443 U.S. at 538, 99 S.Ct. at 2979 (emphasis added) (citations omitted). See also Swann, 402 U.S. at 26, 91 S.Ct. at 1281 (burden on school board to establish current racial composition of schools not the result of their present or past actions). The school district must show that no causal connection exists between past and present segregation, not merely that it did not intend to cause current segregation. The causal link between prior and current segregation is not snapped by the absence of discriminatory intent alone, or even by a *862firm commitment to desegregation, where it is not accompanied by action that in fact produces a unified school district. Id.
The dissent clearly has misconstrued the parties’ burdens. The emphasis throughout the dissent is on the necessity of plaintiffs proving a current condition of intentional segregation before any burden shifts to the school district. Although the revised dissent has removed a number of its prior references to “intentional” when discussing “current condition of segregation,” and contends that it has not misapplied the burden, the tenor of the current version of the dissent belies this assertion. See Dissent at 892 (“In a former de jure system, plaintiffs may establish the prima facie case by proving that there is a current condition of intentional segregation”) (emphasis added); id. at 904 (“a plaintiff may satisfy his initial burden of proof by relying upon evidence of ‘recent and remote intentionally segregative actions’ of the school board”) (emphasis added); id. at 917 (“This court misses the mark if it is implying that intent is not relevant in this ease.”); id. at 920 (“intent, whether proven by direct or circumstantial evidence, remains an essential element.”); id. at 920-21 (“It is clear that to sustain the presumption, the plaintiffs were required to prove ‘a current condition of segregation from intentional state action.’ ”) (emphasis added); id. at 952 (“The requirement of a current condition of segregation resulting from past or present segregative intent (as opposed to segregation resulting from voluntary demographic change), prior to the operation of the presumption, allows a school system to redeem itself.”) See also Dowell v. Board of Educ., 890 F.2d 1483, 1519 (10th Cir.1989) (Baldoek, J., dissenting) (“To understand the problems with this court's approach, it is necessary to understand how the burden of proof is allocated before a school district becomes unitary. In a system that was statutorily or officially dual, plaintiffs may establish the prima facie case by proving that there is a current condition of intentional segregation; that the de jure system or its vestiges remain or were reestablished in part of the school system.”) (emphasis added).
On the contrary, however, once a school system has been declared unconstitutional because of the existence of de jure segregation, “[u]ntil the ... School System achieves unitary status, official action that has the effect of perpetuating or reestablishing a dual school system violates the defendants’ duty to desegregate.” Pitts v. Freeman, 755 F.2d 1423, 1427 (11th Cir.1985) (emphasis in original); see also, e.g., Morgan v. Nucci, 831 F.2d 313, 329 (1st Cir.1987) (“[T]he fact that a particular school policy or program may be ‘racially neutral,’ in that it no longer reflects discriminatory animus, does not prove that the effects of prior discrimination have been purged”). Even the dissent in Dayton II recognized that “the affirmative duty renders any discussion of segregative intent after 1954 gratuitous.” 443 U.S. at 542-43, 99 S.Ct. at 2982-83.
The dissent falls into this error by viewing this case as one of initial liability. The cases the dissent relies on for its statements about plaintiffs’ need to prove intentional segregation are all cases describing a plaintiff’s duty to establish de jure segregation in the first instance. See, e.g., Columbus Bd. of Educ. v. Penick, 443 U.S. 449, 99 S.Ct. 2941, 61 L.Ed.2d 666 (1979) (dissent at 892, 903-904, 911, 917); Keyes (dissent at 892, 903-904, 919, 920-21). This is not such a case. As in Swann, plaintiffs here are alleging that a school system which has already been declared unconstitutional in earlier years has never fulfilled its affirmative duty to eradicate the effects of that segregation. Under these circumstances, plaintiffs need only show that current racial disparities exist, not that such disparities are the result of current intentional segregation on the part of the school board. See Swann, 402 U.S. at 26, 91 S.Ct. at 1281. This is not a liability case in which plaintiffs must prove that the school board has committed unconstitutional acts; that was established in 1954.
*863Where a plaintiff has established segregation in the past and the present, it is “entitled to the presumption that current disparities are causally related to prior segregation, and the burden of proving otherwise rests on the defendants.”27 School Board of the City of Richmond v. Baliles, 829 F.2d 1308, 1311 (4th Cir.1987); see also Dayton II, 443 U.S. at 536, 99 S.Ct. at 2978 (systemwide nature of de jure schools in 1954 “furnished prima facie proof that current segregation ... was caused at least in part by prior intentionally segregative official acts [and] judgment for the plaintiffs was authorized and required absent sufficient countervailing evidence by the defendant school officials.”); Keyes, 413 U.S. at 211 n. 17, 93 S.Ct. at 2699 n. 17 (after de jure segregation is established, “the burden becomes the school authorities’ to show that the current segregation is in no way the result of those past segregative actions.”); Vaughns v. Board of Educ., 758 F.2d 983, 991 (4th Cir.1985) (same). This presumption ensures that subconscious racial discrimination does not perpetuate the denial of equal protection to our nation’s school children.28 A focus on provable intent alone would deny a remedy to too many Americans.
The dissent asserts that under this view of the law, liability in this case would be frozen as of 1954 and “we merely would decree a remedy based upon the failure of the school board to have absolute racial balance at every school and much of this court’s opinion ... would be superfluous.” Dissent at 894. Of course, liability is not frozen; it just remains until the school board, not the plaintiffs, bears the burden of proving that current racial disparities in the school system are not the result of the prior segregated school system.
Contrary to the district court’s apparent conclusion, see 671 F.Supp. at 1297, remoteness in time does not make past intentional acts less intentional. See Dayton II, 443 U.S. at 535-36, 99 S.Ct. at 2977-78; Keyes v. School Dist. No. 1, 413 U.S. 189, 210-11, 93 S.Ct. 2686, 2698-99, 37 L.Ed.2d 548 (1973). The passage of time merely presents an opportunity for a school district to show that the presumptive relationship between the de jure system and the current system is so attenuated that there is no causal connection. See id. at 211, 93 S.Ct. at 2699.
What the school district has done to integrate is crucial in determining whether the causal link between the prior segregation and the current disparities has been severed. The district may carry its burden by showing that it has acted affirmatively to desegregate. Absent such proof, the court must presume that current segregation is the result of prior intentional state action. A showing that the school district has not promoted segregation and *864has allowed desegregation to take place where natural forces worked to that end is insufficient.
The ultimate test of what the school district has done is its effectiveness, most significantly its effectiveness in eliminating the separation of white and minority children.29 While a district is not always required to choose the most desegregative alternative when it selects a particular option,30 the result of the sum of the choices made by the district must be to desegregate the system to the maximum possible extent.31 Furthermore, the school district may “not ... take any action that would impede the process of disestablishing the dual system and its effects.” Dayton II, 443 U.S. at 538, 99 S.Ct. at 2979.
One choice frequently made by school districts, and the one made in Topeka, is to use a neighborhood school plan as the basis for student assignment. Neighborhood schools are a deeply rooted and valuable part of American education.32 To the extent that neighborhoods are themselves segregated, however, such plans tend to prolong the existence of segregation in schools.33 Thus, they must be carefully scrutinized. They are not “per se adequate to meet the remedial responsibilities of local boards.” Davis, 402 U.S. at 37, 91 S.Ct. at 1292; see United States v. Board of Educ., Indep. School Dist. No. 1, Tulsa County, 429 F.2d 1253 (10th Cir.1970).34
Neighborhood school plans must be both neutrally administered and effective. A plan that is administered in a scrupulously neutral manner but is not effective in producing greater racial balance does not fulfill the affirmative duty to desegregate.35 It is equally important that a plan’s neutrality be more than surface-deep. We have specifically held that when minorities are concentrated in certain areas of the city, neighborhood school plans may be wholly insufficient to fulfill the district’s affirmative duty to eliminate the vestiges of segregation. Tulsa County, 429 F.2d at 1258-59. Even when neighborhood school plans hold the promise of being effective, courts must recognize that the school district’s choices on such questions as where to locate new schools, which schools to close, how to react to overcrowding or underutilization, and what transfer policy to offer, all have obvious impact on the school *865attendance boundaries the district can draw under a neighborhood school plan.36 If these choices are not made with an eye toward desegregation, a neighborhood school plan may “further lock the school system into a mold of separation of races.” Swann, 402 U.S. at 21, 91 S.Ct. at 1278. Ultimately, whether the use of a neighborhood school plan in a particular case is consistent with a school district’s duty to desegregate turns on whether the “school authorities [have made] every effort to achieve the greatest possible degree of actual desegregation taking into account the practicalities of the situation.” Davis, 402 U.S. at 33, 91 S.Ct. at 1290. Contrary to the dissent’s suggestion, dissent at 897, a school district which chose a neighborhood school system in 1954 is not insulated from a later contention that neighborhood schools did not result in desegregation.37 “A school system is not ... automatically desegregated when a constitutionally acceptable plan is adopted and implemented, for the remnants of discrimination are not readily eradicated.” Ross v. Houston Indep. Sch. Dist., 699 F.2d 218, 225 (5th Cir.1983).38
Actions the school district has not taken are also relevant in considering what the district has done. A school district which has not made use of such classic segre-gative techniques as gerrymandering, discriminatory transfer policies, and optional attendance zones is more likely to have fulfilled its duty to desegregate than a district that has done so.39 Similarly, a school district that has made use of the various techniques available to encourage voluntary desegregation is more likely to have fulfilled its duty than one that has not.40 Such techniques may include, for example, the establishment of magnet schools and vigorous official encouragement of desegregative transfers.
Finally, objective proof of the school district’s intent must be considered. How a district lobbies its patrons and government agencies on issues that affect desegregation, whether it seeks and then heeds the desegregation recommendations of others, and the cooperativeness of the district in complying with court orders, for example, bear on the manner in which the district has shaped the current conditions in the school district.41
C. Maximum Practicable Desegregation
What more can and should be done, if anything, is the final component in a determination of unitary status.42 Essen*866tially, a defendant must demonstrate that it has done everything feasible. Courts must assess the school district’s achievements with an eye to the possible and practical, but they must not let longstanding racism blur their ultimate focus on the ideal.43
In most unitariness cases, the school district has been implementing a court-approved desegregation plan under active court supervision. The question is usually whether closer adherence to the plan is practical or whether the plan has achieved its objectives.44 The district court in such cases has been intimately involved with the process of desegregation and is well aware of the obstacles it faces. The court can thus make an informed judgment on the possibilities of further desegregation. Where the school district has complied with the desegregation plan to the best of its ability, and has done what can be done in spite of the obstacles in its way, it is reasonable to conclude that no further desegregation is feasible.45
The present case is one of those rare ones in which the unitariness determination is not directly tied to the execution of a particular desegregation plan. In such a case, the consideration of whether further desegregation is practicable must include the obstacles that are likely to stand in its way, and whether they may be circumvented without imperiling students’ health or the educational process. See Swann, 402 U.S. at 30-31, 91 S.Ct. at 1282-83. Where there are no significant barriers to desegregation, or such barriers as exist may be overcome without undue hardship, further desegregation is practicable. See id. at 28, 91 S.Ct. at 1282 (mere awkwardness or inconvenience is no barrier to carrying out desegregation plan).
In sum, when a school system was previously de jure, a plaintiff bears the burden of showing that there is a current condition of segregation. It may do so by proving the existence of racially identifiable schools. The school district must then show that such segregation has no causal connection with the prior de jure segregation, and that the district has in fact carried out the maximum desegregation practicable for that district. We now apply these legal principles to Topeka.
V.
THE FINDING OF UNITARINESS
Because Topeka’s schools formerly operated under a system of de jure segregation, “[t]he board’s continuing obligation ... [has been] ‘to come forward with a plan that promises realistically to work ... now, ... until it is clear that state-imposed segregation has been completely removed.’ ” Columbus Board of Educ. v. Penick, 443 U.S. 449, 459, 99 S.Ct. 2941, 2947, 61 L.Ed.2d 666 (1979). Prior to this case, no court had pronounced the Topeka school system unitary; hence, this duty never dissipated. The district court concluded, however, that the effects of de jure segregation have been eliminated in Topeka. On appeal, plaintiffs attack this determination.
A. Burden of Proof
Plaintiffs argue initially that the district court improperly required them to prove intentional discriminatory conduct on the part of the school district over the course of the decades instead of according them the benefit of a presumption that current segregation stems from the prior de jure system. Plaintiffs quote a number of sentences from the district court’s opinion as support for their argument that the court placed on them the burden of proof on intent. Brief for Plaintiffs-Appellants at 27.46 The court itself expressed some con*867fusion as to the proper burden of proof. 671 F.Supp. at 1295. We have considered both these citations and the tenor of the district court’s opinion as a whole, and we are convinced that the court focused too greatly on the school district’s lack of discriminatory intent. Although the percentage of minority students in Topeka is lower than in other cities involved in desegregation cases and consequently the statistics alone do not appear as egregious, we are persuaded that this overemphasis on the school district’s intent led the court to make the same errors as did the district court in Dayton II. It failed “to apply the appropriate presumption and burden-shifting principles of law.” Brinkman v. Gilligan, 583 F.2d 243, 251 (6th Cir.1978), aff'd sub nom. Dayton Board of Education v. Brinkman, 443 U.S. 526, 99 S.Ct. 2971, 61 L.Ed.2d 720 (1979).47
The district court made the following findings: that the neighborhood school attendance boundaries drawn in 1955 had the effect of maintaining segregation; that the construction of new schools since that time had the effect of “promot[ing] racial separation”; that the reassignment of students from previous de jure schools to adjacent schools with higher-than-average percentages of minority students had the effect of increasing those percentages; and that the assignment of faculty had the effect of placing minority faculty disproportionately at schools with higher-than-average minority student percentages. 671 F.Supp. at 1300, 1301, 1304-05. It is clear from the court’s other findings that the school district’s use of space additions, its siting of Topeka West high school, its drawing of attendance boundaries, and its failure to adopt various reorganization plans did not further the process of desegregation. Id. at 1298-1301, 1308-09. Nevertheless, the court’s discussion of most of these aspects of Topeka’s history ends with the conclusion that because these actions were not taken with the intent to discriminate and were consistent with a “race-neutral” neighborhood school plan, they did not promote segregation,48 The court evidently believed that if these two criteria, i.e., no intent to discriminate and consistency with a race-neutral neighborhood school plan, were met, the school district’s actions would pass constitutional muster. The Supreme Court has made clear, however, that a “race-neutral” neighborhood school policy is not sufficient where it fails to remedy de jure segregation:
“Absent a constitutional violation there would be no basis for judicially ordering assignment of students on a racial basis. All things being equal, with no history of discrimination, it might well be desirable to assign pupils to schools nearest their homes. But all things are not equal in a system that has been deliberately constructed and maintained to enforce racial segregation. The remedy for such segregation may be administratively awkward, inconvenient, and even bizarre in some situations and may impose burdens on some; but all awkwardness and inconvenience cannot be avoided in the interim *868period when remedial adjustments are being made to eliminate the dual school systems.
"... ‘Racially neutral’ assignment plans proposed by school authorities to a district court may be inadequate; such plans may fail to counteract the continuing effects of past school segregation resulting from discriminatory location of school sites or distortion of school size in order to achieve or maintain an artificial racial separation. When school authorities present a district court with a ‘loaded game board,’ affirmative action in the form of remedial altering of attendance zones is proper to achieve truly non-discriminatory assignments. In short, an assignment plan is not acceptable simply because it appears to be neutral.”
Swann, 402 U.S. at 28, 91 S.Ct. at 1282 (emphasis added).49
While we agree with the district court’s findings that the current school administration is not presently acting with discriminatory intent — indeed, there is evidence that the present school board has some commitment to desegregation — we are persuaded that the court failed adequately to weigh the conduct of the school district for the past thirty years, and the current effects of that conduct. The court erred by limiting the school district’s burden merely to showing that it had nondiscriminatory reasons for acting as it did. As thirty years of desegregation law have made clear, the Constitution requires more than ceasing to promote segregation. See part IV supra. “[T]he measure of the post-Brown I conduct of a school board under an unsatisfied duty to liquidate a dual system is the effectiveness, not the purpose, of the actions in decreasing or increasing the segregation caused by the dual system.” Dayton II, 443 U.S. at 538, 99 S.Ct. at 2979. A lack of intent to discriminate is therefore insufficient. “In short, [a student] assignment plan is not acceptable simply because it appears to be neutral.” Swann, 402 U.S. at 28, 91 S.Ct. at 1282. Mere adherence to a race-neutral but ineffective neighborhood school plan is insufficient. In general, any course of action that fails to provide meaningful assurance of prompt and effective disestablishment of a dual system is unacceptable. Wright v. Council of City of Emporia, 407 U.S. 451, 460, 92 S.Ct. 2196, 2202, 33 L.Ed.2d 51 (1972). The district court did not heed this mandate. While it did find that the school district had taken some actively desegregative actions, we are convinced that the court’s overall conclusion as to unitariness was fatally infected by the inadequacy of the burden of proof standard to which it held the school district.
B. The Evidence
In order to assess the district court’s finding of unitariness under the appropriate burden of proof and the general principles we have outlined, we turn to a more specific review of the record. As a general matter, it is important to note that much of the record evidence consists of statistics and other undisputed facts.50 Our differ-*869enees with the district court lie mainly in how the essentially undisputed facts are assessed in light of the school district’s mandate to dismantle the segregated school system. We believe that the district court’s finding of unitariness is flawed by the undue deference it gave to the school district’s neighborhood school policy and by the court’s failure to give proper weight to its own findings that certain actions and omissions by the school district had a seg-regative effect.51 Cf. Diaz v. San Jose Unified School Dist., 733 F.2d 660, 674 (9th Cir.1984) (en banc) (“In making its determination in respect to the Board’s intent, the district court erred in failing to give weight to the cumulative impact of the evidence”; district court’s findings that defendants acted without segregative intent held clearly erroneous).
1. Current Condition of Segregation in Topeka
The district court found that “there are disparities in the racial makeup of various schools’ enrollments,” and that “[pjlaintiffs have demonstrated that in general there are a greater than average number of minority faculty and staff in schools with a greater than average number of minority students.” 671 F.Supp. at 1295, 1304. Like most courts, however, the district court did not discuss separately the issues of current segregation and the causal con-neetion between that segregation and the prior de jure segregation. As we have pointed out, the simplest and most compelling evidence of segregation is the presence of predominantly one-race schools. In a system such as Topeka’s, however, in which the minority student population is relatively small, there may be a number of primarily white schools even though minority students are spread through a significant number of other schools. In such a system, it is the concentration of minority students that is usually the hallmark of discrimination.52 Because the significance of mostly white schools is therefore not necessarily as great in a mostly white system as it would be in a system with a heavy minority population, we focus on the broader form of racial identifiability discussed in part IV A above. In support of their argument that there is currently segregation in Topeka, plaintiffs point primarily to student assignment, and faculty and staff assignment.53 We consider each in turn, and then together.
a. Student Assignment
Each of the experts who testified at trial used a different standard for determining whether a school was racially identifiable in student assignment. Plaintiffs’ main experts, Mr. Lamson and Dr. Foster, each used standards that took the percentage of black or minority students actually enrolled in the elementary or secondary schools *870(26% in 1985), and then added and subtracted some number to obtain a range within which they did not consider schools to be racially identifiable on the basis of student assignment alone. Their methods differ to some extent, but for 1985 either method leads to a range of 11-41% (26% plus or minus 15%).54 The school district’s primary expert on this issue, Dr. Armor, used an absolute rather than a relative standard. In his view, desegregated schools should optimally have 20-50% minority students, regardless of the percentage of minority students in the system. Dr. Armor also allowed a variance, which resulted in a range of acceptability of 10-60%.55
As plaintiffs point out, under any of these methods there are schools in Topeka that are racially identifiable by student assignment. Even under the most generous of these numerical standards, proposed by the school board’s expert, there are six elementary and three secondary schools that are racially identifiable by student assignment.
Notwithstanding the dissent’s view that the statistics regarding the percentages of minorities in the racially overbalanced schools do not add up to a current condition of segregation, we believe the figures speak for themselves. See supra at 856-58. In this connection, the charts in the dissent showing the racial inventories over the years at Belvoir, Hudson, Highland Park North, Lafayette, Quinton Heights, and Lowman Hill speak more eloquently about the current condition of segregation in Topeka than words ever could. See dissent at 936 n. 43; 938 n. 44; 939 n. 45; 939-40 n. 46; 941-42 n. 47; 946 n. 51.
b. Faculty and Staff Assignment
To determine racial identifiability by faculty/staff assignment, plaintiffs again used a standard based on the actual percentage of minority employees and a range of a few percentage points above and below that number. The school district contested the accuracy of plaintiffs’ standard but presented no alternative one. We do not adopt plaintiffs’ standard, but instead evaluate the data on its face.56
In 1985, the percentage of minority faculty/ staff in the Topeka school system was 11.2% for the elementary schools and 12.65% for the secondary schools. In the *871elementary schools, the percentage of minority faculty/staff at individual schools ranged from 0% to 33.3%. Nine schools had less than 5% minority faculty/staff, and two had more than 25%.57 In the secondary schools, the percentage ranged from 2.5% to 24.7%. One school had less than 5% minority faeulty/staff, an additional three schools had less than 10%, and one had more than 20%.58
Faculty/staff data have been kept only since 1973 and, except for 1981, that data does not distinguish between faculty and staff. Rec., ex. vol. IV, at 263-68. Faculty/ staff includes managerial personnel at both the school and district level, teacher aides, clerical/secretarial employees, skilled and technical employees, and service workers, as well as teachers and other professional staff. The distinction between facul-
*872ty and staff is particularly relevant because the percentage of minority employees has always been lower than the minority student population, and has fallen steadily at the elementary level over the period such data was kept. Moreover, minorities are represented more heavily in staff positions than in faculty positions. In 1985, for example, district-wide statistics showed that 11.3% of elementary teachers and 8.0% of secondary teachers were minorities, while 19% of teacher aides and 20% of service workers were minorities. Rec., vol. IV, at 268. Any one faculty/staff person listed at any one school is thus more likely to be a teacher aide or service worker than a teacher.
We recognize that the small number of faculty and staff at any one school means that the presence or absence of one minority employee may have a considerable effect on the school’s minority percentage. Nevertheless, we see no obviously neutral reason why McClure elementary school has no minority employees among its 25 faculty/staff and Topeka West high school has 3 among 120, while Avondale East elementary school has 10 minority faculty/staff out of a total of 31 employees and Robinson middle school has 12 out of 49.
c. Factors considered together
Because faculty/staff assignment is largely within the control of the school district, it is a potent tool for demonstrating that the district does or does not itself identify certain schools as white or minority. It also provides an opportunity for undoing some of the harm of segregated student assignments, because both white and minority students may benefit from the presence of minority role models. See Washington v. Seattle School District, No. 1, 458 U.S. 457, 472, 102 S.Ct. 3187, 3196, 73 L.Ed.2d 896 (1982) (“white as well as Negro children benefit from exposure to ethnic and racial diversity in the classroom”). Conversely, if the district disproportionately assigns minority faculty/staff to those schools with the highest percentages of minority students, the district is in effect reinforcing the identification of particular schools as white or minority. This practice of disproportionate assignments also reinforces the irrational notion that minority teachers are inferior and not fit to teach white children.
In Topeka, although the correlation is not completely uniform, see 671 F.Supp. at 1305, there is a clear pattern of assigning minority faculty/staff in a manner that reflects minority student assignment. This correlation is fatal to the school district’s effort to show a lack of current segregation. Both student assignment and faculty/staff assignment can be expected to vary from school to school, the former because of population distribution, and the latter, to a lesser extent, because of differing teacher credentials. When they vary together, as they do in Topeka, leading to schools that are noticeably more white or more minority in both students and faculty, it is difficult to posit a neutral explanation.59 The school district has not attempted to provide one.
*873Moreover, when we look beyond the numbers, we find that the schools that are marked as white or minority by their students and faculty/staff are also so marked by their geography, the residential population in their attendance areas, and by their history. Of the six racially identifiable elementary schools detected by Dr. Armor’s method, five are now and always have been attended almost exclusively by white students. They are located on the western and northwestern edges of the school district, areas with mostly white populations.60 The same is true of the three secondary schools. See infra part V B(2)(c)(vii). The one remaining elementary school, Belvoir, is located on the eastern edge of the school district. The area has long been inhabited by a significant minority population, and the school’s student population is now and has been for over twenty years more than half minority. See infra part V B(2)(c)(i). Finally, the correlation between student assignment and faculty/staff assignment is not a one-year fluke. The same correlation has existed throughout the course of this litigation. See infra part V B(2)(b)(ii). Considering all of these factors together, there is sufficient evidence to support plaintiffs’ contention that there is a current condition of segregation in Topeka.61
*8742. The Causal Link Between De jure Segregation and the Current Condition of Segregation
Brown I established that the Topeka school system was one of de jure segregation. Because there is a current condition of segregation, we turn our attention to the causal link between these two conditions of segregation, which must be assessed in light of the burden and factors set out in part IV B, supra. We are convinced that the school district failed to meet its burden of showing the absence of this link. This failure, which the district court did not see because it failed to impose on defendants the proper burden of proof, is the key to our reversal.
Timing is central to an assessment of the Topeka school district’s actions. After what was described as a “good faith” beginning, Brown, 139 F.Supp. at 470, the course the school district followed in the late 1950s and early 1960s may fairly be characterized as segregative. The decade from 1956 to 1966 is important because it established a framework from which the school district subsequently deviated very little.62 A period of quiescence then followed, during which the system was simply administered as it stood. It was not until HEW threatened to withhold federal funds in the mid-1970s that the school district undertook some positive action to desegregate its schools. After that brief flurry of action, the school district again turned its attention elsewhere, but to its good fortune the oft-maligned forces of demography began to work in its favor. Two things are apparent from the record. First, Topeka has largely acted as if it had fulfilléd its duty to desegregate at the conclusion of the four-step plan implemented in the 1950s. Second, although Topeka’s schools have in fact become less segregated in the last decade, this lessening of segregation is due in part to forces beyond the control of the school district. Moreover, those desegregative actions which the school board did undertake were primarily the result of pressure from the federal government, or were instituted after this lawsuit was filed.63 Although its record is better than that of many other school districts, Topeka has engaged in voluntary desegregation with little enthusiasm.
a. The general pattern in each decade i. The mid-50s to mid-60s
This period was one of significant change in the Topeka school district. Most notably, the district expanded greatly with several city-imposed annexations at the end of the 1950s, the beginning of a spurt in population growth and shift to the newly annexed areas, and the school district’s consequent opening of new schools in this outer white part of Topeka. As the white population moved outwards, the inner city population became increasingly heavily minority, and inner city schools were closed.
The mid-60s found the Topeka school system still heavily segregated. While the numeric polarization between schools had *875decreased to some extent systemwide,64 and minority students were somewhat less concentrated,65 the number of schools serving primarily white children had increased. Geographic polarization also increased, as a result of the building of so many primarily white schools on the outer edges of the district. Plaintiffs introduced evidence tending to show that the school district’s use of portable classrooms and optional attendance zones served to maintain segregation by concentrating students of one race at certain schools. The school district’s expert, Dr. Clark, conceded that his study of changes in attendance zone population because of changed attendance boundaries led him to conclude that one such boundary change might have had seg-regative effects not explainable solely by demographic shifts.66
We have no doubt that during this period the school district in fact maintained and perhaps promoted a segregated system by current standards. Moreover, the system that existed after the wave of school openings and closings ended, i.e., the location of schools and the race of their students, formed the basis for the current elementary system. Therefore, while the school district should not be judged primarily by actions now twenty or more years in the past, neither can those actions be ignored.
ii. Mid-60s to Mid-70s
This period was one of quiescence in the school district. Enrollment in the Topeka school system peaked in 1969, substantially ending the need for new school buildings. Outer Topeka continued to grow in white population, particularly in the western part of the city. Minority population began to spread out of its highly concentrated central areas into eastern Topeka.
The only significant change in the school system at the end of this period was that the number of virtually all-white schools dropped. At the elementary level, 19 out of a total of 35 elementary schools were virtually one-race in 1966; 13 out of a total of 34 were virtually one-race in 1974.67 The change at the secondary level was even less: from 6 out of 14 schools in 1966, to 5 out of 15 schools in 1974. This change took place primarily in schools on the outskirts of the southeastern part of Topeka, the area into which minorities were spreading. The school district’s conduct during this period can thus be summarized as letting demographic forces work without interference or encouragement. This also means that schools already heavily minority were allowed to increase in minority population.68 It is apparent that while Topeka did not promote a segregated school system during this period, it maintained the system then in effect notwithstanding its affirmative duty to eliminate segregated schools.
iii. Mid-70s to the present
In 1974, the HEW compliance action began the third phase of the Topeka school system since Brown I. At the elementary level, HEW cited unequal facilities for minority and white children as well as “stu-
*876dent racial compositions not consonant with a unitary plan.” Five elementary schools were specifically listed as having “substantially disproportionate minority student compositions clearly the result of a former dual pattern of operation.” At the secondary level, HEW found that the junior high schools attended by most minority students were inferior in facilities to those schools attended largely by white students. In addition, the district’s transfer plan was criticized.69 The Board denied that the district was in noncompliance and obtained an injunction against further HEW administrative proceedings. Nevertheless, the Board agreed to take “administrative steps to assure a more perfect unitary school district.” 70 It developed and implemented two plans largely approved by HEW.
These plans had some success. At the end of the reorganization, there were no 90 + % minority elementary schools, the attendance boundaries of two 90 + % white schools had been redrawn so that they were no longer one-race, and a third one-race white school had been closed. At the secondary level, two heavily minority and one primarily white junior high school had been closed, although the minority population of one other junior high school had risen significantly as a result of the reorganization. The school district’s desegregation indices for elementary schools dropped significantly between 1975 and 1981.71
Some of this decline, however, was the result of the movement of minorities into the western part of the school district. During the same six-year period, three additional elementary schools rose above the 10% minority level solely because of this demographic change.72 This movement has continued. In 1985, two additional elementary schools were just barely no longer one-race for that reason.73
The defendants’ experts, Dr. Clark, see, e.g., rec., vol. XI, at 2338-39, and Dr. Armour, see, e.g., rec., vol. XIII, at 2579, the district court, see 671 F.Supp. at 1302-03, 1310, and the dissent, see, e.g., dissent at 937, 938, 939, 941, all mistakenly believe that the changing demographics which resulted in unfavorable racial balances at various schools excuses the school board from achieving more acceptable racial balances. This view is clearly wrong where, as here, the school system is under an affirmative duty to dismantle a dual school system. “Until a unitary system is created, a school system is not absolved from this duty by reason of demographic changes.” Vaughns v. Board of Educ., 758 F.2d at 988; see also, Lee v. Macon County Bd. of Educ., 616 F.2d 805, 809-10 (5th Cir.1980); cf. Tasby v. Wright, 713 F.2d at 91 n. 3, 93 (school districts’ white population was 58.2% when lawsuit filed in 1970 and dropped to 29.53% by 1981; school district not yet unitary). It is only
“[w]hen state officials have not only made good-faith efforts to eliminate the vestiges of segregation, but have actually achieved a school system clean of every residue of past official discrimination, [that] immutable geographic factors and post-desegregation demographic changes that prevent the homogenation of all student bodies do not bar judicial recognition that the school system is unitary.”
Ross v. Houston Indep. Sch. Dist., 699 F.2d at 225. The school district’s excuse for the racial imbalance which exists today in Topeka’s schools is that shifting residential patterns caused the problem. That is an unacceptable excuse when the school *877district has never fully dismantled the effects of the prior de jure segregation.
Other changes that took place in Topeka from the mid-70s to the date of trial are as follows: elections to the Board were changed in 1976 to ensure that all parts of the city were fairly represented; the Board has now, and has had for some time, significant minority representation; in 1981, the district abandoned a slightly segregative majority-to-minority transfer policy for a transfer policy that is slightly desegrega-tive; in 1984, while this trial was pending, a black superintendent was appointed; and, five days before the beginning of trial, the district adopted an explicit policy against discrimination in faculty and staff assignment.74 These changes are deliberate and voluntary actions on the part of the school district, and constitute evidence that the district is, as it claims, now committed to desegregation.75
There is no doubt that the Topeka school system has improved dramatically in the last ten years as far as desegregation is concerned. However, the system was equally undoubtedly in need of such improvement. The question now is whether these changes broke the causal link between the system that existed prior to them and the current system. Based on our review of the record, we have a definite and firm conviction that they did not. Approximately one quarter of Topeka’s minority elementary students still attend three schools more than 50% minority, two of which have fallen into this category since at least 1966. One-race white schools have existed throughout the relevant period; many remain, and those that are no longer within the 90 + % white range are still heavily white schools. Most of the reorganization took place in the center of the school district; the schools on the periphery—the ones most clearly marked as minority or white schools—were largely unaffected. The distribution of faculty/staff has improved, but in 1985 there was still one school with no minority faculty/staff and eight schools with only one full- or part-time minority faculty/staff person, who may or may not be a teacher or other professional, and these schools are those same schools that have lacked minority faculty/staff since the 1960s. This is a system improved, but it is still the same system.
b. Individual Factors
Green v. County School Board, 391 U.S. 430, 435, 88 S.Ct. 1689, 1693, 20 L.Ed.2d 716 (1968), identified faculty, staff transportation, extracurricular affairs, and facilities as facets of school operations that must be considered in determining racial identifiability of schools. Swann added student attendance, school siting, opening and closing, and the drawing of attendance boundaries as factors to be weighed. Swann, 402 U.S. at 25-29, 91 S.Ct. at 1280-83. The district court made detailed findings as to most of these. We discuss below those factors that most clearly demonstrate the continuing causal link between past and present segregation.
i. Student assignment over time
We begin by considering the subsequent history of the schools in existence at the time of Brown II. No real pattern emerges. Seventeen elementary schools have more or less maintained the same racial identity for the last twenty years, while nine have changed. For the most part, the major shifts in student attendance mandated by the school district have had relatively little effect, although there has been some *878improvement in the last ten years.76 Nevertheless, the record convinces us that the causal connection between the Topeka school system in 1955 and the same system in 1985 has not been broken. The school district has unquestionably had the opportunity to draw up and execute a scheme designed to lead to comprehensive desegregation. Moreover, Topeka does not have the kind of geographical barriers that have sometimes frustrated desegration plans.77 Its minority population is spread over half the city, not concentrated in hard-to-reach enclaves.78 It has altered the attendance boundaries of almost every school and closed many of them; it has in place a slightly desegregative transfer plan that potentially could be significant; and it has apparently faced little or no racially-motivated opposition from the community.79 Nonetheless, there is, we repeat, no pattern to the changes over the years. There should have been one. We simply see no evidence that Topeka dedicated itself to desegregation prior to the reopening of this case.80 Although former members of the Board testified that they did not vote for plans with segregative effects, they did not testify that they regularly took desegregation concerns into account.81 The reaction of at least one Board member to the white flight of the 1950s and 1960s was “What could we do? We can’t make people not move, no way.” 82 After hearing the testimony of the Board members, the district court commented, “I am not interested in just any intentional acts that would hinder integration. I am also wondering if there is any action taken by this school board and past school boards to promote that integration, rather than hinder.... [A]re we going to get to this ... ?” Rec., vol. XII, at 2534. This perceptive question focused on precisely what is missing from the school district’s evidence. By no stretch of imagination can the school district’s conduct be characterized as acting “with all deliberate speed” to “convert to a unitary system in which racial discrimination would be eliminated root and branch.”83
We do not ignore the changes made to the school system in the late 1970s. In our view, however, the effect of these changes was to make a system that was highly segregated as to student assignment into a system that is still segregated, although somewhat less so.
*879ii. Faculty/staff assignment over time
Plaintiffs do not dispute the district court’s finding that the hiring of minority faculty and staff is not now discriminatory. 671 F.Supp. at 1304. They do challenge the court’s conclusion with respect to the assignment of faculty and staff. They point out that the court in fact found that the assignment of faculty/staff is disproportionate, and they argue that this disproportion, and its consistency over time, is “one of the classic indices of a segregated school system.” Brief for Plaintiffs-Appellants at 38. They criticize the district court for having unduly minimized the import of this evidence of segregation.
For example, while it is true that all of the district’s schools have been within 10% of the average percentage minority faculty/staff at least once in the last ten years, 671 F.Supp. at 1305, that statement is misleading considering that the percentage of minority faculty/staff in the system as a whole has ranged only from 9.9% to 16% in the last ten years at the elementary level and is currently about 11%. Thus, in 1975, when the average percentage was 9.9%, schools with no minority faculty or staff were still technically within 10% of the average. The 10% once-in-ten-years standard includes both MeEachron, which had no minority faculty/staff for nine of the 13 years for which such statistics have been kept and in the last three years has had only one minority faculty/ staff member out of approximately 25, and Lowman Hill, which has never had less than 15% and has once had more than 30% minority faculty/staff during the same time period. Furthermore, while it is true, as the district court noted, that the percentage of minority faculty/staff assignment has not invariably tracked the percentage of minority students, there is nevertheless a distinct pattern of correlation. McClure, for example, which has had more than 5% minority students only in 1979 and 1985, has had no minority faculty/staff since 1979, while Avondale East, which has generally had minority student percentages in the 30s and 40s has also generally had minority faculty/staff percentages in the 20s and 30s. Finally, because our data does not distinguish between faculty and staff, and because minority employees are more heavily represented as staff than as faculty, it is quite possible that the one minority faculty/staff person at MeEachron, for instance, is a janitorial staff person or other service worker rather than a teacher or other professional role model.
The long-standing pattern of imbalances coupled with the tracking of student assignment percentages is strong evidence of a dual system because, as pointed out above, faculty/staff assignment is far less difficult to adjust than a factor such as student assignment. While the ratio of minority faculty/staff in schools with different levels of minority students has improved over the twelve years that records have been kept, that improvement alone does not alter the pattern. It merely makes it less dramatic.
There is one area in which the Topeka school system is clearly desegregated. Minorities are well-represented, indeed statistically over-represented, at the managerial level. Several principals are minorities, as is the current superintendent. This is both laudable in itself and an indication that whatever bias lingers in the assignment of faculty/staff is not deliberate. Nevertheless, although there has been improvement in this area, racial disparity does in fact linger because the school district has not consciously addressed the problem.84 Inertia has thus led to the maintenance, albeit in less striking form, of a system that has kept white faculty at primarily white schools and minority faculty at predominantly minority schools.
*880iii. Attendance boundaries and how (not) changed
Attendance boundaries determine the neighborhoods from which neighborhood schools draw their students. They can be used as an important tool in either imposing or undoing segregation under an ostensibly race-neutral neighborhood school policy. We do not disagree with the district court’s finding that attendance boundaries have not been drawn in recent years with racial animus. In the past, however, the school district drew boundaries with significantly segregative effects, although for the most part it avoided obvious gerrymandering. Many of the schools affected by segregative boundary changes were subsequently closed, and other schools lost their status as predominantly minority schools bordered by schools with much lower percentages of minority students when minorities moved into the residential areas of the bordering schools.
A number of existing schools do have peculiar-looking attendance boundaries. Randolph has an arm jutting out to the east. Stout and Quinton Heights both include areas to the north and east some distance away from the school building. In the case of Randolph and Stout, the extensions have a desegregative effect. The reverse is true of Quinton Heights’ current boundaries. See infra Part V B(2)(c)(v). As far as the record reflects, the current boundaries of Randolph and Stout are the only examples that Topeka has manipulated attendance boundaries in order to desegregate. Again, when it altered attendance boundaries the school board failed to use the opportunity in most instances to further desegregation, despite its affirmative duty to do so.
As the district court acknowledged, the most serious example of segregative attendance boundaries is Lowman Hill. Low-man Hill’s attendance area is a more or less square area with the school located in the center. Plaintiffs point out that this area includes, and has included for the last 25 years, two areas of heavy minority population. These areas are now part of a general central Topeka minority population, but were relatively isolated when Lowman Hill’s boundaries were first drawn to include them. Within the last ten years, three elementary schools bordering the Lowman Hill area, one with a minority student percentage significantly lower than Lowman Hill’s, have closed.85 The closings did not affect the minority student percentage at Lowman Hill. It remains a school with a heavy minority percentage bordered by schools with much lower minority percentages. It also remains the school whose attendance boundaries include the only significant group of minority population in the northwestern quarter of the school district. See infra Part VI B(2)(c)(vi).
The history of Lafayette is similar although not as extreme. The border between Lafayette and State Street, the school to Lafayette’s north, has functioned as the boundary between the mostly white area served by State Street and the school to its north and the heavily minority area served by Lafayette and the schools to its south and southwest. Lafayette has long had twice the minority population of State Street; thus, since 1966, Lafayette’s minority population has ranged from 52% to 69% while State Street’s has ranged from 27% to 34%.86 See infra Part VI B(2)(c)(iv).
While the boundaries of Loman Hill and Lafayette are consistent with a neighborhood school policy, as held by the district court, the original decisions to draw them to include the black population of the area had a segregative effect which the school district has never undone. The school district may not draw segregative attendance boundaries, maintain them for decades, and then take shelter behind the neutrality of its recent inaction. See Swann, 402 U.S. at 21, 91 S.Ct. at 1278.
*881iv. Location of new schools
All of the elementary schools and all but one of the secondary schools opened by the school district since Brown II were built in the white outer part of the district. Robinson middle school, which serves the central part of Topeka, is the sole exception. Furthermore, many of these schools were built near to the edges of the district. Plaintiffs criticize both the location of these schools and the fact that they opened at all. With respect to the location, they argue that Topeka deliberately placed new schools as far as possible from minority residential areas in order to polarize the district into white outer and minority inner schools. With respect to the new schools themselves, they argue that Topeka should have taken advantage of the underutilized inner schools; that is, that Topeka should have voluntarily bused white children from outer Topeka to the inner city schools instead of putting up new buildings.
Almost all of the building in question took place in the late 1950s and early 1960s, before the Supreme Court made clear that prior de jure school districts must do more than operate a neutral system. Green, 391 U.S. 430, 88 S.Ct. 1689. Moreover, the city was annexing new territory during that time, a matter beyond the control of the school district, and it is not entirely unreasonable for the district to build schools “where the children are.” Nevertheless, the erection of so many schools devoted to educating white children had a significant segregative effect on the district. See infra Part V B(2)(a)(i). While the building of new schools to accommodate an influx of white children arguably had a legitimate basis at the time, the school district did nothing to diminish the segregative effect of those schools, even after it became clear that school districts under an unfulfilled duty to desegregate must do more than operate a facially neutral neighborhood school system. As we have said, neighborhood school plans are permissible if they are effective in actually reducing segregation. School openings and locations are a crucial variable in such plans. Swann, 402 U.S. at 20-21, 91 S.Ct. at 1278-79. In Topeka, school openings under the neighborhood school plan in fact worked to increase segregation. A school district conscious of its constitutional duty would have attempted to counteract that effect. See Diaz v. San Jose Unified School District, 733 F.2d 660, 667-69 (9th Cir.1984) (en banc), cert. denied, 471 U.S. 1065, 105 S.Ct. 2140, 85 L.Ed.2d 497 (1985). Topeka did nothing.
v. School closings
In contrast to school openings, the majority of school closings took place in the central and middle parts of Topeka. Like school openings, this pattern was dictated largely by the population shift to the outer part of the school district. Plaintiffs do not contend that schools were closed for any other reason, or that the impact of the closings was borne primarily by minority students. They focus instead on the effect of school closings, primarily on the opportunity closings provided for more equitably distributing students to other schools. They contend that school closings of highly segregated schools were not desegregative in practice because the school district simply reassigned the students at those schools en masse to a nearby school which then took over the segregated status of the school closed. The evidence supports this contention for the 1950s and 1960s. Although most of the receiving schools remained schools with a majority of white students, plaintiffs argue that these reassignments served to identify the schools to which the school district would subsequently channel minority students. School closings in the 1970s and beyond had less segregated effects. In this decade, the effects of closings have been mixed. A few school closings were desegregative. These later closings have not completely undone the effect of the earlier closings, however. On the whole, most school closings in the 1970s cannot be said to be either segre-gative or desegregative.
In sum, although the hiring of minority faculty and staff became nondiscriminatory, and even included significant representation of minorities in management, the district nevertheless maintained racial im*882balance in the assignment of faculty and staff. In addition, the location of new schools intensified the system’s dual nature. The drawing of attendance boundaries and other means of reassigning students were rarely used to desegregate. The closing of schools is the most significant desegregative device Topeka utilized. Even here, it is only in a few cases that a closing can unqualifiedly be characterized as desegregative. As we stated earlier, Topeka notably failed to focus its efforts on the complete eradication of the effects of prior segregation.
c. Individual schools
The dissent spends a great deal of time attacking our analyses of individual schools. The main point of that attack is that the increasing minority percentages for these schools resulted from demographics, not from school board actions. As we have previously stated, however, a school district is not absolved by demographics from its affirmative duty to desegregate before it achieves unitariness. See supra at 876-77.
Plaintiffs describe the history of the school district as one of targeting certain schools as minority. According to plaintiffs, while the identity of those schools shifts as population moves, the district as a whole is always polarized into schools with clear racial identities. We therefore consider what the district did and did not do with respect to individual schools.
i.Belvoir elementary school
Belvoir is and has been since the 1960s a school with one of the highest minority populations in the district. The school was annexed in the late 1950s and lies on the eastern edge of the current district boundary. Two federal housing projects were built within its attendance area in 1963 and 1965. Since 1966, its student population has never been less than 50% minority.87 It is located in an area of minority population that was bordered by white populations to the north and south and a minority population to the west when the school was annexed. The bordering areas are now all areas with substantial minority populations.
Belvoir’s boundaries were once altered to take in part of the attendance areas of a school that was closed. That closing did have a modest desegregative effect at Bel-voir, but it did not change Belvoir’s status as the school with the highest minority percentage in the system. Although it is no longer surrounded by schools with markedly lower minority percentages, it remains an east-side minority school as opposed to the west-side white schools. It cannot be said that the neighborhood school plan has been effective in desegregating Belvoir.
ii.Hudson elementary school
The school district built Hudson in 1963 in newly annexed territory. It is on the southeastern edge of the district and opened as a primarily white school. While the passage of time has dissipated Hudson’s identity as a white school, this change is due solely to the spread of minorities into its attendance area, not to any action by the school district. The school was 46.55% minority in 1985-76. Rec., ex. vol. IY at 171.
iii.Highland Park North elementary school
Highland Park North was annexed to the school district in the late 1950s. Its percentage of minority students has risen significantly over the years: 21.6% in 1966, 42.1% in 1975, and 57.9% in 1985. This rise was paralleled by a shift in the surrounding residential population from primarily white to heavily minority. This shift is characteristic of the eastern part of the city, where Highland Park North is located. The only significant action the school district took with respect to Highland Park North was to assign it Parkdale students when that school closed. This reassignment increased Highland Park North’s minority percentage at a time when it was already high.
*883iv.Lafayette elementary school
Lafayette was a white school in the Topeka school system at the time of Brown I. It was 13.7% black at the end of the four-step plan in 1956. Federally subsidized housing projects were built in Lafayette’s area in 1961 and 1962. By 1966, the school was 54.5% minority (26.1% black). The school was then a transition school between Parkdale elementary school to the southwest, with 85% minority students, and Sumner, State Street, and Rice elementary schools to the north and east, with 12.3%, 27.5% and 3% minority students, respectively. Belvoir, to the south, had approximately the same percentage of minority students. Lafayette has remained 55-65% minority (25-50% black). In 1985, Lafayette was 56.8% minority (41.4% black). It remains a transition school between schools with higher minority percentages to the south and lower percentages to the north and west.
The closing of three schools affected Lafayette’s boundaries. We have insufficient information to evaluate one closing, in the 1960s, and the other two closings had negligible effects on Lafayette’s minority percentages. Under its neighborhood school plan, the school district simply maintained Lafayette as a school with disproportionate minority percentages.
v.Quinton Heights elementary school
Quinton Heights was a white school at the central southern part of the school district in the 1950s. At the end of the four-step desegregation plan it was 7.3% black. In 1959, Pierce, an annexed black school, and in 1964, Van Burén, a formerly all-white school, were closed and some of their students sent to Quinton Heights. Quinton Heights’ attendance boundary was extended to the north as a result.88 By 1966, its minority student percentage was 36.4%. It was bordered by heavily minority Monroe elementary school to the northeast and otherwise by primarily white schools, including schools to the south annexed in the late 1950s.
Plaintiffs characterize the reassignments in the 1960s as part of a process of focusing Quinton Heights as a school for black students. In 1966, Quinton Heights’ minority student percentage was seventh highest out of thirty-five elementary schools, and its minority population bordered two schools with much lower minority percentages (Central Park, with 15.5%, and Polk, with 11.5%). Indeed, on the maps available to us, it is quite noticeable that Polk’s boundaries in 1970 correspond closely to a patch of white population surrounded by minority population assigned to other schools, including Quinton Heights.89
Two adjacent schools were closed during the 1970s. Each time, Quinton Heights’ boundary was extended northwards. It is thus clear that Quinton Heights’ current boundaries are the result of continued manipulation by the school district. They extend far into minority areas to the north and east of the school building and cross many natural barriers.90 Quinton Heights is a “neighborhood” school only in the sense that some of its students do live in the immediate area of the school building. Most do not. Quinton Heights is entirely the product of the school district’s actions; as such, the district may not disclaim responsibility for the makeup of the students who attend it.
vi.Lowman Hill elementary school
Lowman Hill was a white school in the pre-Brown Topeka school system. It had 17.4% black students at the end of the four-step plan in 1956. During the next decade a number of small adjustments were made to optional attendance zones around the school. Plaintiffs describe these adjustments as a process of drawing *884Lowman Hill’s boundaries to include as many black and as few white students as possible. In 1959, an all-black school, Buchanan, was closed and its students sent to Lowman Hill. By 1966, Lowman Hill was 49.6% minority. The minority percentage at the five surrounding schools ranged from .2% to 15.6%. It is noticeable on the maps in the record that in 1970 Lowman Hill’s boundaries included the only two significant areas of minority population in that part of Topeka.91
Lowman Hill’s minority student percentage has ranged in the 40s since that time, dipping briefly into the high 30s in the early 1980s. It was not affected in any significant way by the reassignment of students due to the closing of neighboring Clay elementary school in 1975 and Polk in 1979; while its boundaries expanded somewhat, the student population it acquired was apparently of the same makeup as its existing population. Thus, the history of Lowman Hill is that the boundaries drawn in the 1950s and 1960s largely remain, as do their effects. Lowman Hill still has a significantly higher minority student percentage than its neighbors, although the disparity between them is less (in 1985, Lowman Hill was 41.9% minority while its four neighbors ranged from 7.7% to 31.5%). The school appears to have been designed and maintained as a school with a concentration of minority students. At best, it has been neglected in recent decades. See infra Part V B(2)(b)(iii).
vii. Predominantly white schools
Eight elementary schools were at time of trial and have been during the course of this litigation primarily white schools.92 Even in 1985, only three of these schools crossed the 10% minimum that the school district’s expert proposed, and did so only just barely.93 The schools are in or near the western part of the district. Similarly, at the time of trial two middle schools and one high school — one-third of the secondary school system — were, and had been since they opened, primarily white schools. They too are in the western part of Topeka.
Plaintiffs focus on Topeka West High School as the quintessential example of deliberate channeling of white students to schools in the western part of the school district. The school was opened in 1961 six blocks from the western edge of the school district, the part of the district most distant from centers of minority population. Consequently, it was then and remains now a school with more than 90% white students, while the percentage of minority students at the other two high schools has been higher than average. Little evidence was introduced on the other white schools that has not already been discussed elsewhere. See infra Part V B(2)(b)(iv). To summarize, these schools were either annexed or built by the school district in the late 1950s and early 1960s. They opened as white schools with white faculty/staff and have largely remained so. The school district offered no evidence that it had ever attempted to eliminate these schools’ identity as schools for white children.
3. Maximum Practicable Desegregation in Topeka
We look at last away from what has been done and what should have been done to what can still be done. The feasibility of further measures was not a focus of the case, and there was little evidence on the question. The district court explicitly found that “[a]t any time, more could have been done to achieve racial balance in the schools.” Other courts that have found school systems to be unitary have specifically found that additional desegregation was impracticable.94 Similarly, courts have *885refused to find school systems unitary where greater integration is achievable.95
It is apparent from this record that further desegregation in Topeka is not constrained by geographic obstacles as in some other cities.96 Furthermore, out of the panoply of desegregation tools that have been developed, Topeka at present apparently uses only one: its minimally effective majority-to-minority transfer plan. We have no reason to think that Topeka has exhausted the repertoire available for desegregating schools. For example, it has failed to try magnet schools or to aggressively encourage voluntary transfers to improve racial balance.97
Indeed, there is every reason to believe that further desegregation in Topeka is feasible. The present commitment of the school board and district administration, the already widespread distribution of minorities, and the expected continued movement of minorities into western Topeka should make the course of further desegregation relatively smooth.
It was suggested at trial that further desegregation is, in effect, not worth the trouble. Under questioning from counsel for the State Board of Education, plaintiffs’ expert Foster admitted that in the 1985 school year the movement of only 265 elementary students, or 3% of all elementary students, and 77 secondary students, or 1.2% of all secondary students, would have been sufficient to bring all of Topeka’s schools into plaintiffs’ 15 + /% range. Dr. Foster pointed out that this figure was arrived at by simply removing students from schools outside the 15 + /% range, without analyzing where they would go or which students would take their place. The actual movement of students would thus be more complex and burdensome than the numbers suggest. Dr. Foster also pointed out that such an approach would move schools now just beyond the 15% range to just within it, an improvement he did not find greatly desegregative as it would preserve the existence of schools differing by 30% in their minority percentage and thus still marked as minority or *886white schools.98 A similar analysis was performed with respect to faculty/staff.99
The exercise above is illuminating in a number of respects. First, it indicates again that the desegregation plan which must be developed may not be too burdensome, a circumstance for which all can be thankful. Second, it reminds us of the dangers of focusing on numbers alone. Although numbers are usually the focus of desegregation plans, racial balance in itself is not the goal. The goal is to wipe out the effects of prior segregation. Setting target ranges is one means to that goal. It is clearly not the only one. Furthermore, as we have emphasized, schools are not racially identifiable by student and faculty assignment alone.
The lack of evidence or any attempt to argue that further desegregation is impracticable is perhaps the largest flaw in the school district’s case. When the law requires the maximum desegregation reasonably feasible, and evidence suggests that more can be done, unitariness must await the implementation of those further steps. When it can be said that Topeka can do no more to eradicate the effects of past segregation and segregative acts, the Topeka school system may be declared legally unitary.
C. Conclusion on Unitariness
After reviewing the record, we have concluded that there is a current condition of segregation in Topeka. Contrary to the district court, we are convinced that this condition is causally connected to the prior de jure system of segregation. This causal connection for the most part does not stem from active promotion of segregation. Topeka has generally heeded the prohibition against this form of discrimination.
Our basic disagreement with the district court is centered around the fact that the duty to desegregate requires more. What Topeka did not do is actively strive to dismantle the system that existed. It opened and closed schools with little or no thought given to the effect of such actions on segregation, and observed the segregative or desegregative effects of such actions with indifference. Even those school board decisions that had some desegregative effect were not carried out in such a manner as to disestablish the dual system. For example, rather than integrate the former de jure black schools, Topeka gradually closed them. Their students were then reassigned virtually en masse to other schools, which then took on the same minority-school identity.
The shifting distribution of Topekans throughout the city sometimes hindered, sometimes aided, the cause of desegregation, but the school district sought neither to reduce the impact of the former nor to encourage the effects of the latter. It simply administered the system, with due regard for economic efficiency and the quality of education in the district, but without a commitment to undoing the segregated structure of that education. Consequently, it permitted Topeka’s schools to fall into three categories: schools that are now and always have been white, schools that are now and long have been heavily minority, and others. As we have noted, the mere existence of racially identifiable schools does not violate the Constitution. Where prior de jure segregation exists, however, we are convinced that permitting white schools and minority schools to remain racially identifiable as such without significant efforts to the contrary is in effect to permit the continuation of a dual system of education. Based on the evidence set out above, we have a definite and firm conviction that the district court erred when it found the Topeka school system to be unitary. Cf. Pitts v. Freeman, 755 F.2d 1423 (11th Cir.1985); Diaz, 733 F.2d 660; Adams v. United States, 620 F.2d 1277 (8th Cir.1980) (en banc); Brinkman v. Gil*887ligan, 583 F.2d 243 (6th Cir.1978), aff'd sub nom. Dayton Board of Education v. Brinkman, 443 U.S. 526, 99 S.Ct. 2971, 61 L.Ed.2d 720 (1979).
VI.
TITLE VI
Plaintiffs argue that the school district violated both section 601 of Title VI of the Civil Rights Act of 1964, 42 U.S.C. § 2000d, and 34 C.F.R. § 100.3(b). Section 601 forbids programs and activities receiving federal funds from discriminating on the basis of race, among other factors,100 and 34 C.F.R. § 100 contains regulations implementing Title VI. The Topeka school system receives federal funds.
The Supreme Court held in Regents of the University of California v. Bakke, 438 U.S. 265, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978), that discrimination under Title VI is to be measured by the constitutional standard of the Equal Protection Clause. Id. at 284-87, 98 S.Ct. at 2745-47. Because we have concluded that Topeka’s school system is not unitary under constitutional standards, we necessarily reverse the district court’s holding that the school district has not violated Title VI.
VII.
STATE DEFENDANTS
The state of Kansas intervened as a defendant in Brown I to defend the constitutionality of the Kansas law permitting segregation. It was not again actively involved in the litigation of the case until May 1980, when the Governor and members of the State Board of Education were joined as defendants for the purpose of granting injunctive relief.101 The district court granted summary judgment before trial for the Governor on both the constitutional and Title VI claims. After trial, it held that the State Board of Education was not responsible for the racial conditions in the Topeka school district. We affirm both of these rulings.
The supreme executive power of the state of Kansas is vested in its Governor, who “shall be responsible for the enforcement of the laws of this state.” Kan. Const, art. I, § 3. One of the Governor’s powers is that of reorganizing state executive agencies. “[Cjonstitutionally delegated functions of state officers and state boards” are exempt from this reorganization power. Id. § 6(a). The State Board of Education is such an exempt state board. Id. art. 6, § 2; State ex rel. Miller v. Board of Education, 212 Kan. 482, 511 P.2d 705, 709 (1973). Because the State Board of Education is the Kansas organ of government responsible for education, these provisions limit the Governor’s executive power over education to that embodied in the power to enforce the laws of Kansas. Plaintiffs have not suggested that any Kansas law is now implicated in this litigation. The Governor is not vested by the provisions cited above with the power to enforce federal law, and plaintiffs have not shown any other source of such power. The district court was correct in concluding that the Governor is not able, under state law, to comply with such injunctive relief as may be ordered in this case. It was therefore proper to dismiss him from the case.
The correct disposition of the case against the members of the State Board of Education is more difficult to resolve. Kansas, like many other states, assigns the primary responsibility for the day-to-day management of public schools to local school authorities. See Kan. Const, art. 6, § 5. At the time of Brown I, the Kansas Constitution provided for a Superintendent of Public Instruction with limited authority. In particular, this officer did not have the *888authority to exercise general supervision of the public schools. Miller, 511 P.2d at 708. Apparently, the only method used by the state to encourage segregation was the permission granted by the law struck down in Brown I for segregation below the high school level. The Kansas legislature promptly repealed this statute in its next full session after Brown I, and has since enacted civil and criminal laws against racial discrimination. See Kan.Stat.Ann. §§ 21-4003, and §§ 44-1001 et. seq. This history is in contrast to the actions of other states that have been held liable for promoting segregation.102 Moreover, the state was never held expressly liable for the actual segregation in Topeka’s schools.
It is not necessary for us to decide whether requiring more of the state at this point would be consistent with the principle that a remedy may be no broader than the scope of the violation found, because plaintiffs have not shown that the State Board of Education has the power to act as they would have it act. The Kansas Constitution provides both for the State Board of Education and for locally elected school boards. The State Board is given the power of “general supervision” of the local boards. Kan. Const, art. 6, §§ 2, 5. The Kansas Supreme Court has defined this power as “something more than to advise but something less than to control.” Miller, 511 P.2d at 713. Among the powers of the State Board is accreditation of schools. Plaintiffs argue that the State Board could have withheld accreditation of segregated schools, citing evidence that the Board is empowered to withhold accreditation on this basis.103 The evidence refers, however, to hypothetical situations such as an official policy of not hiring black teachers or complete segregation of students. This does not establish that the State Board had the power to deny accreditation under the facts of this case. In addition, exhibits in the record show that the State Board was aware that there was actual segregation in some Kansas cities, and drew up a number of proposals to provide advisory assistance to these cities in desegregating.104 These proposals are consistent with the State Board’s argument that its role is primarily advisory in the area of desegregation.
Kansas statutes authorize local school boards to appoint employees to serve at the board’s pleasure, set attendance boundaries, open and close schools, and adopt all necessary rules and regulations. Kan.Stat. Ann. §§ 72-8202e, 72-8212, 72-8213. These tools are the primary means of desegregating. No evidence suggests that the State Board exercises control over these tools. Plaintiffs argue that after Brown I some agency of the state had a duty to ensure that desegregation occurred. We are not persuaded that any authority other than the Topeka school district had then or has now the power to desegregate Topeka’s schools.
VIII.
CONCLUSION
In concluding, we begin by stressing what we do not have in Topeka. We do not *889have a school system or a community actively resisting desegregation. Nor, on our record, has Topeka ever interposed the kind of obstacles to desegregation found in other cities. In recent times, the school district has won national recognition for its innovative work on its curriculum and has been honored with various awards for the excellence of its schools. In short, the Topeka school district is actively engaged in improving the education of its students.
This active engagement has largely been directed at concerns other than desegregation, however. Once the four-step plan approved by the district court in the 1950s was implemented, Topeka did not until very recent times, on our record, give serious consideration to the question of whether the duty imposed by Brown II had been fulfilled, save under the urging of HEW. Although that urging led to distinct improvements, for the most part the Topeka school district has exercised a form of benign neglect. The duty imposed by the Constitution, and articulated in numerous cases by the highest court in this land, requires more.
We reverse the finding of the district court that the Topeka school system is legally unitary. We remand the case to the district court for the formation of an appropriate remedy.
. Statement by William Lamson during trial. Rec., vol. II, at 162-63. Mr. Lamson has done demographic studies in desegregation cases since 1970. His services as an expert have been utilized by the Department of Justice, the Department of H.E.W., the N.A.A.C.P., the A.C. L.U., and various school boards. Rec., supp. vol. 12 pi. ex. 25. In 1975, the defendant Board of Education hired him to put together a ten-year forecast of student enrollments. Rec., vol. II, at 103.
The district court echoed Mr. Lamson’s view about the nature of segregation: "[RJacially conscious student assignment with the goal of racial balance has been approved as a remedy in desegregation litigation. This is because school segregation has an inertia which often cannot be countered by a purely neutral force." Brown v. Board of Educ., 671 F.Supp. 1290, 1297 (D.Kan.1987) (citation omitted) (emphasis added).
. Section 601 states:
"No person in the United States shall, on the ground of race, color, or national origin be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance." 42 U.S.C. § 2000d (1982). The Topeka school district received federal funds through the Kansas State Department of Education.
. Linda Brown, a child named plaintiff in the original suit, is now the mother of two intervening child plaintiffs.
. Rec., ex. supp. vol. 13, pi. ex. 155 B. The record on appeal consists of pleadings, transcripts, and exhibits. We cite them respectively as "Rec., doc. #,” "Rec., vol. and "Rec., ex. vol.
. Rec., ex. vol. IV, at 54-56.
.Lyman elementary school had been dean-nexed in 1967. Rec., vol. Ill, at 281.
.Rec., ex. vol. IV, at 134-38.
.Percentage of Minority Students In Topeka Elementary Schools In 1985
School Per Cent School Per Cent
Avondale East 44.1 Lundgren 15.8
Avondale West 16.6 McCarter 9.2
Belvoir 61.9 McClure 7.2
Bishop 10.5 McEachron 10.3
Crestview 8.9 Potwin
9.4 Quincy 20.5
Highland Park Central 35.1 Quinton Heights 49.4
Highland Park North 57.9 Randolph 14.8
Highland Park South 28 Shaner 20.7
Hudson 46.55 State Street
56.8 Stout 26.8
Linn 29.4 Sumner 31.5
Lowman Hill 41.9 Whitson 10.2
27.2% of all elementary students in 1985 were minorities. Source: Rec., ex. vol. IV, at 170-74.
.Rec., vol. Ill, at 306-07. Plaintiffs’ expert Lamson used the figures for black rather than minority students in his analysis and testimony. Where we repeat his figures, we therefore refer to black students and white students. We also refer to black students when we discuss pre-1966 numbers, as it is only in that year that figures begin to be available for minority students generally. Otherwise we refer to minority students. See Keyes v. School Dist. No. 1, 413 U.S. 189, 197, 93 S.Ct. 2686, 2691, 37 L.Ed.2d 548 (1973). The parties are in agreement that the difference in analysis between black students and minority students is not significant in this case. Rec., vol. IV, at 409-10; rec., vol. V, at 598, 602-03.
. Rec., vol. VII, at 1050.
. Id. at 1051.
. Rec., ex. vol. II, at 56-57.
. Percentage of Minority Students In Topeka Secondary Schools in 1985
Middle Schools Per Cent High Schools Per Cent
Chase 33.4 Highland Park 33.6
Eisenhower 48.7 Topeka High 30.9
French 6.2 Topeka West 7.9
Jardine 17.3
Landon * 9.3
Robinson 28.5
The percentage of minority students in all middle schools was 26.9%, while the minority percentage at the high school level was 23.8%.
Source: Rec., ex. vol. IV, at 175-77.
Landon is now closed.
. See, e.g., Riddick v. School Bd. of the City of Norfolk, 784 F.2d 521, 533 (4th Cir.), cert. denied, 479 U.S. 938, 107 S.Ct. 420, 93 L.Ed.2d 370 (1986); United States v. Texas Educ. Agency, 647 F.2d 504, 506 (5th Cir.1981), cert. denied, 454 U.S. 1143, 102 S.Ct. 1002, 71 L.Ed.2d 295 (1982); cf. Dayton Bd. of Educ. v. Brinkman, 443 U.S. 526, 534 & n. 8, 99 S.Ct. 2971, 2977 & n. 8, 61 L.Ed.2d 720 (1979) (whether school district is intentionally operating a dual school system is a question of fact).
. The Fifth Circuit defines unitariness as " ‘a district in which schools are not identifiable by race and students and faculty are assigned in a manner that eliminates the vestiges of past segregation.’ ” Monteilh v. St. Landry Parish School Bd., 848 F.2d 625, 629 (5th Cir.1988) (quoting United States v. Lawrence County School Dist., 799 F.2d 1031, 1034 (5th Cir.1986)).
. Cf. Morgan v. Nucci, 831 F.2d 313, 319 (1st Cir.1987) (considering number of one-race or racially identifiable schools, good faith on the part of the school district, and maximum practicable desegregation); Ross v. Houston Indep. School Dist., 699 F.2d 218, 227 (5th Cir.1983) (considering conditions in district, accomplishments to date, and feasibility of further measures).
. The Supreme Court desegregation cases involved school systems in which the degree of segregation was sufficiently great that the parties did not seriously dispute on appeal that the plaintiffs had satisfied their burden on this issue. See Dayton II, 443 U.S. at 529, 99 S.Ct. at 2975 (Dayton public schools "highly segregated by race"); Wright v. Council of City of Emporia, 407 U.S. 451, 455, 92 S.Ct. 2196, 2199, 33 L.Ed.2d 51 (1972) (complete segregation); Swann, 402 U.S. at 24, 91 S.Ct. at 1280 (no *860challenge to finding of prior dual system); Green v. County School Bd. of Educ., 391 U.S. 430, 435, 88 S.Ct. 1689, 1692, 20 L.Ed.2d 716 (1968) (complete racial identification of schools). The issue was potentially more significant in recent circuit cases in which a school district had been under court order for some time and many of the vestiges of prior de jure segregation had been eliminated. Even in these more recent cases, however, no clear standard has been articulated. See Morgan, 831 F.2d at 319-21 (considering number of one-race schools as part of unitariness determination); Price v. Denison Indep. School Dist., 694 F.2d 334, 347-68 (5th Cir.1982) (discussing need to consider various factors in determining whether constitutionally violative condition of segregation exists).
. Given modern urban demography and geography, one-race schools may well have evolved for reasons beyond school board control. See, e.g., Calhoun v. Cook, 522 F.2d 717, 719 (5th Cir.1975). Thus, where a school system consists of de facto one-race schools, rather than de jure, the system is not unconstitutional.
. See Morgan, 831 F.2d at 320 (listing standards ranging from 70% to 90% and declining to decide whether 80% or 90% is more appropriate for Boston); Tasby v. Wright, 713 F.2d 90, 91 n. 2, 97 n. 10 (5th Cir.1983) (90% standard for one-race schools; 75% standard for predominantly one-race schools). Swann did not define the term “one-race school," presumably because two-thirds of Charlotte-Mecklenburg’s black students attended schools that were 99 + % black. See Swann, 402 U.S. at 7, 91 S.Ct. at 1271.
. See Morgan, 831 F.2d at 320 n. 7 (rejecting 75% standard in district 72% black); Castaneda v. Pickard, 781 F.2d 456, 461 (5th Cir.1986) (school 97.88% Mexican-American not a vestige of discrimination in district 88% Mexican-American); Ross, 699 F.2d at 220, 226 (affirming finding of unitariness for district 80% minority although 57 out of 226 schools were 90 + % one-race); Price, 694 F.2d at 336, 339-40 (schools not necessarily racially identifiable in district 88% white although 7 out of 8 elementary schools 90 + % white); Calhoun, 522 F.2d at 718-19 (85% black district unitary although more than 60% of schools all or substantially all black).
. See Dayton II, 443 U.S. at 529 n. 1, 99 S.Ct. at 2975, n. 1; Milliken v. Bradley, 418 U.S. 717, 726, 94 S.Ct. 3112, 3118, 41 L.Ed.2d 1069 (1974); Ross, 699 F.2d at 226; Lee v. Macon County Bd. of Educ., 616 F.2d 805, 808-09 (5th Cir.1980).
. See Morgan, 831 F.2d at 320; Tasby, 713 F.2d at 91 n. 2; Ross, 699 F.2d at 226; Price, 694 F.2d at 364; Stout v. Jefferson County Board of Education, 537 F.2d 800 at 802 (5th Cir.1976).
. See Texas Educ. Agency, 647 F.2d at 508; cf. Lee v. Tuscaloosa City School System, 576 F.2d 39 (5th Cir.) (per curiam); United States v. Board of Educ. of Valdosta, Ga., 576 F.2d 37 (5th Cir.) (per curiam) cert. denied, 439 U.S. 1007, 99 S.Ct. 622, 58 L.Ed.2d 684 (1978); Carr v. Montgomery County Bd. of Educ., 377 F.Supp. 1123, 1134 (M.D.Ala.1974), aff’d, 511 F.2d 1374 (5th Cir.) (per curiam), cert. denied, 423 U.S. 986, 96 S.Ct. 394, 46 L.Ed.2d 303 (1975).
. See Milliken v. Bradley, 433 U.S. 267, 280 n. 14, 97 S.Ct. 2749, 2757 n. 14, 53 L.Ed.2d 745 (1977); Pasadena City Bd. of Educ. v. Spangler, 477 U.S. 424, 434, 96 S.Ct. 2697, 2703, 49 L.Ed.2d 599 (1976); Swann, 402 U.S. at 24-26, 91 S.Ct. at 1280-81.
. See Morgan, 831 F.2d at 320 (noting difficulty of further desegregating schools located in geographically isolated or heavily black sections of Boston); United States v. Lawrence County School Dist., 799 F.2d 1031, 1043-44, 1047 (5th Cir.1986) (considering demography and geography in reversing trial court’s refusal to order new student assignment plan). Price, 694 F.2d at 347-68 (authoritatively demonstrating that degree of racial balance is only one of many factors to be considered); Stout, 537 F.2d 800 (affirming remedy leaving three schools one-race because of geographic isolation and barriers); cf. Carr, 377 F.Supp. at 1141 (criticizing formulas for determining racial balance as "highly artificial” and severely disruptive).
. See Keyes v. School Dist. No. 1, 413 U.S. 189, 196, 93 S.Ct. 2686, 2691, 37 L.Ed.2d 548 (1973) (what is a segregated school depends on facts of the particular case; faculty and staff percentages and community and administrative attitudes as well as racial composition of student body are relevant); Price, 694 F.2d at 347-68; Lawrence County School Dist., 799 F.2d at 1039-40 (looking to student and faculty percentages and history and location of school).
. The dissent mistakenly refers to this as the "Keyes presumption.” Dissent at 904, 919-20, 920-21. The Keyes presumption states "that a finding of intentional segregative school board actions in a meaningful portion of a school system ... creates a presumption that other segregated schooling within the system is not adventitious.” Keyes, 413 U.S. at 208, 93 S.Ct. at 2697. This is a presumption that is used to establish initial de jure segregation in a case where no statutorily mandated segregation exists. The presumption referenced in the instant case applies only after either statutorily-mandated or policy-based de jure segregation has been established. See id. at 211 n. 17, 93 S.Ct. at 2699 n. 17 (Keyes court making this distinction); see abo Tasby v. Wright, 713 F.2d 90, 94 (5th Cir.1983) (district court relied on wrong presumption). The dissent’s misunderstanding of this difference has an obvious and critical impact on its assessment of this record. See dissent at 920-21 (“It is clear that to sustain the presumption, the plaintiffs were required to prove 'a current condition of segregation from intentional state action'”); id. at 951 ("The court simply has jumped the gun in invoking the Keyes presumption to arrive at the preliminary and critical conclusion that there is a current condition of segregation in Topeka.”); see abo id. at 951-52 (dire predictions if Keyes presumption is required at the outset). The presumption required here is not the Keyes presumption but the presumption placed on all school districts previously found liable for operating a de jure system.
. As one commentator has observed, "[A]meri-cans share a historical experience that has resulted in individuals within the culture ubiquitously attaching a significance to race that is irrational and often outside their awareness.” Lawrence, The Id, the Ego, and Equal Protection: Reckoning With Unconscious Racbm, 39 Stan.L.Rev. 317, 327 (1987).
. See Wright v. Council of City of Emporia, 407 U.S. 451, 462, 92 S.Ct. 2196, 2203, 33 L.Ed.2d 51 (1972); Davis, 402 U.S. at 37, 91 S.Ct. at 1292; Swann, 402 U.S. at 25, 91 S.Ct. at 1280.
. See Pitts v. Freeman, 755 F.2d 1423, 1427 (11th Cir.1985).
. See Diaz v. San Jose Unified School Dist., 733 F.2d 660 (9th Cir.1984) (en banc) (castigating school district for consistently choosing more segregative alternatives), cert. denied, 471 U.S. 1065, 105 S.Ct. 2140, 85 L.Ed.2d 497 (1985).
. See 20 U.S.C. § 1701 (1982) (declaring it to be the public policy of the United States that neighborhood schools are the appropriate basis for determining public school assignments); Crawford v. Los Angeles Bd. of Ed., 458 U.S. 527, 537 n. 15, 102 S.Ct. 3211, 3217 n. 15, 73 L.Ed.2d 948 (1982); Diaz, 733 F.2d at 677 (Choy, J., dissenting).
. See Swann, 402 U.S. at 28, 91 S.Ct. at 1282; Diaz, 733 F.2d at 664.
. On remand, the district court in Tulsa County developed a plan to desegregate Tulsa’s schools, which we subsequently affirmed. 459 F.2d 720 (10th Cir.1972). The Supreme Court then summarily reversed our affirmance of the proposed plan and remanded for reconsideration in light of Keyes. 413 U.S. 916, 93 S.Ct. 3048, 37 L.Ed.2d 1038 (1973). We then determined that “the factual premise upon which we based our original decision ha[d] been so materially changed both by lapse of time and the specific and voluntary actions taken by the School Board and the students themselves that our further consideration under the present record would serve no useful purpose.” 492 F.2d 1189 (10th Cir.1974). We remanded to the district court for such further proceedings as might be necessary to bring the school district in conformity with the Keyes mandate. Our original decision overturning the district court's finding of no constitutional violation remains the law of this circuit.
.See Morgan, 831 F.2d at 328-29 (racial neutrality is “unreliable talisman”); Diaz, 733 F.2d at 664 (adherence to neighborhood plan not determinative on question of segregative intent); Adams v. United States, 620 F.2d 1277, 1285-86 (8th Cir.1980) (en banc) (adoption of neighborhood school plan did not fulfill duty to desegregate); cf. Pitts, 755 F.2d at 1426 (mere adoption of desegregation plan insufficient to render a dual system unitary).
. See Columbus Bd. of Educ., 443 U.S. 449, 462 & nn. 9-11, 99 S.Ct. 2941, 2949 & nn. 9-11, 61 L.Ed.2d 666 (1979); Swann, 402 U.S. at 28, 91 S.Ct. at 1282; Diaz, 733 F.2d at 667-71; Tulsa County, 429 F.2d at 1256-57.
. See Swann, 402 U.S. at 21, 28, 91 S.Ct. at 1278, 1282; Adams, 620 F.2d at 1286.
. See also Columbus, 443 U.S. at 459-60, 99 S.Ct. at 2947-48 ("In Swann, it should be recalled, an initial desegregation plan had been entered in 1965 and had been affirmed on appeal. But the case was reopened, and in 1969 the school board was required to come forth with a more effective plan"); United States v. Lawrence County School Dist., 799 F.2d 1031, 1043 (5th Cir.1986) ("the adoption of the plan does not exhaust the power of the court to direct elimination of vestiges of segregation that remain or become apparent only after the plan has been put into place"); Pitts, 755 F.2d at 1426.
. See Adams, 620 F.2d at 1288-91 (intact busing, school site selection, block busing, transfer policy, and segregated faculty assignments); Higgins v. Board of Educ., 508 F.2d 779, 787 (6th Cir.1974) (listing segregative techniques); Tulsa County, 429 F.2d at 1257 (transfer policy).
. See Ross, 699 F.2d at 222, 227; Price, 694 F.2d at 351-53; cf. Diaz, 733 F.2d at 672-73 (criticizing school district for implementing none of desegregation proposals made by citizens’ committee).
. See Columbus, 443 U.S. at 463 n. 12, 99 S.Ct. at 2949 n. 12 (Board refused to seek advice on desegregation or implement recommendations); Morgan, 831 F.2d at 321 (noting cooperation with court orders); Diaz, 733 F.2d at 671-74 (manipulation of committee studying segregation; statements suggesting failure of bond issue would lead to forced busing); Ross, 699 F.2d at 222-23 (school district appointed community task force to develop magnet plan, opposed efforts to disrupt integration plans, and promoted interdistrict transfer).
. See Davis, 402 U.S. at 37, 91 S.Ct. at 1292; Morgan, 831 F.2d at 322-25; Ross, 699 F.2d at 224-25.
. See Morgan, 831 F.2d at 324; Ross, 699 F.2d at 225.
. See Morgan, 831 F.2d at 322-25; Riddick, 784 F.2d at 532-34; Calhoun, 522 F.2d 717.
. See Ross, 699 F.2d at 224 (further remedial efforts would be unreasonable and inadequate); Calhoun v. Cook, 525 F.2d 1203, 1203 (5th Cir.1975) (per curiam) ("It would blink reality ... to hold the Atlanta School System to be nonuni-tary because further integration is theoretically possible and we expressly decline to do so.").
.For example, the district court stated:
"Although, on its face, the construction of schools, particularly on the west side of the district, appears to have promoted racial sepa*867ration, the court does not believe that the district’s school construction policy was intended to maintain or promote segregation.”
671 F.Supp. at 1300 (emphasis added).
. The dissent has fallen into the same error, as we have discussed supra at 862-63 & n. 27. The dissent also asserts that we should have remanded to let the district court make appropriate findings in accordance with the presumption and burden-shifting principles we believe apply. Dissent at 890. "But fact findings made under an erroneous view of the law or in the absence of a defined legal standard are not binding on appeal." Oil, Chemical & Atomic Workers v. Amoco Oil Co., 885 F.2d 697, 710 (10th Cir.1989) (Baldock, J., dissenting and urging reversal of trial court’s fact findings on grounds it applied erroneous view of legal standard).
. See, e.g., 671 F.Supp. at 1298-99 ("the use of space additions was consistent with a race-neutral neighborhood policy.... [I]t has not been shown that space additions were intentionally used to promote segregation....”); id. at 1300 ("The court believes the siting of Topeka West High School was a race-neutral decision.”); id. at 1301 (“The district has consistently applied race-neutral, neighborhood school principles to the demarcation of attendance zones.”); id. at 1309 ("The court does not believe the district’s conduct over thirty years indicates a desire to perpetuate segregation by foregoing opportunities to desegregate schools.’’).
. The dissent recognizes this holding in Swann, dissent at 913, but emphasizes that Swann was a remedial case. As we have previously pointed out, one of the dissent’s major errors is its treatment of this case as if plaintiffs were required to establish an initial case of de jure segregation. This is a remedy case. The question presented is whether the school board has remedied the de jure segregation that existed in 1954. One thing is abundantly clear: Creating neighborhood schools in 1954 did not integrate the Topeka schools. See supra at 856 (Three of the four previously all-black schools remained all-black or virtually so under the new neighborhood school policy.); see also Adams, 620 F.2d at 1281 (St. Louis post -Brown neighborhood school plan did not change segregated nature of the system. “Those pre-Brown white schools located in the black neighborhoods ... turned virtually all black immediately after the plan was implemented.”).
. The dissent argues that it is the district court's job, not ours, to decide which party’s experts are more believable. We, of course, agree. In reaching our conclusion that the district court was clearly erroneous in its determination that the Topeka school system has become unitary, we did not rely on the experts’ opinions but rather on the undisputed statistics about the racial makeup of the various schools and their faculty/staffs, and the basically un-controverted evidence about what the school board did, or failed to do, to meet its affirmative duty to dismantle the segregated system.
. We are similarly convinced that the dissent's review of the record is colored by its erroneous understanding of the burdens of proof and the role of intent. See supra at 862-63 & n. 27.
. The dissent is favorably impressed that only three of 34 Topeka schools currently have majority-minority populations. Dissent at 950. The dissent is similarly impressed that of the eight schools that plaintiffs identified at trial as racially identifiable minority, white students make up 49.3% of the total student population at these schools. Dissent at 936. We are not so favorably impressed. In a school system with a 26% minority student population, we think the number of schools in Topeka that approach or are over the 50% minority mark constitute persuasive evidence that the school system has not met its duty to desegregate.
. Plaintiffs do not contest the district court's finding that the Topeka school system is unitary with respect to facilities, extracurricular activities, curriculum, transportation, and equality of education. See Brown, 671 F.Supp. at 1307-08. There are currently no optional attendance zones, and the district’s transfer policy is a majority-to-minority program. See id. at 1298.
Plaintiffs commissioned a public opinion survey in order to determine whether Topekans perceive some schools as black/minority and others as white, whether they perceive some schools as providing an inferior education, and whether there is a correspondence between the two. While the results of the survey provide some support for plaintiffs’ contentions, the survey was extensively criticized as unreliable by several of the school district’s experts. The district court discussed additional flaws and concluded that the survey was not strong evidence of the existence of segregation. 671 F.Supp. at 1305-06. We see no error in that conclusion. We therefore disregard the survey results.
. The dissent attacks the use of a ± 15% deviation from the district average of minority students to assess the racial identifiability of the schools, while at the same time conceding that "[t]he trial judge seems to have analyzed the plaintiffs' case in part with a ± 15% standard," dissent at 921. In Swann, the Court accepted the district-wide average of 71% white/29% black, with no deviation, as an acceptable starting point for the district court in shaping a remedy. 402 U.S. at 24-25, 91 S.Ct. at 1280-81. Where, as here, the school district has been held liable for imposing an unconstitutional segregated school system, we believe it perfectly reasonable to use a ± 15% deviation in assessing whether the school system ever achieved unitar-iness.
. The school district also offered two indices as a measure of desegregation in Topeka’s schools. Rec., vol. XIII, at 2574-80. The dissimilarity index measures how dissimilar schools are compared to the district's mean. Rec., vol. IV, at 558. The exposure index is a measure of the potential for interracial contact. The indices are measures of system-wide desegregation, however; they say nothing about the racial balance in individual schools. Rec., vol. IV, at 554-57; rec., vol. XIII, at 2581. The dissent places great reliance on these indices, pointing out that they have shown significant improvement since 1954. Dissent at 923-25. Defendants’ expert Dr. Armour agreed, however, that the indices are only summary measures, rec., vol. XIII, at 2623-24, and conceded that it is possible to have a school system with a good exposure index that nonetheless contains an all-black school, id. at 2653-54. Therefore, even if a school system has relatively good rates on these indices, it is still necessary to look at the racial statistics at individual schools to get a true picture.
.The range used by plaintiffs was very narrow, and it was extremely difficult for any school to fall within it. In 1985, for example, only four of the twenty-six elementary schools were not racially identifiable under plaintiffs’ standard. While it is appropriate to use a harsher standard for analyzing faculty assignments than student assignments, since the school district may assign faculty as it sees fit, plaintiffs' standard is simply too difficult to meet in this case. We note, moreover, that the school district is limited to some extent in its ability to assign faculty because different teachers are certified in different fields. This is particularly true at the secondary level.
. Minority Faculty/Staff In Topeka's Elementary Schools, 1985-86
School Total Minority Per Cent
Avondale East 31.4 10.45 33.3
Avondale West 25.8 1.0 3.9
Belvoir 25.95 3.85 14.8
Bishop 23.8 .4 1.7
Crestview 31.65 3.4 10.7
4.8 Gage 21.0 o
11.3 Highland Park Central 34.6 vo
19.2 Highland Park North 28.6 In
18.5 Highland Park South 29.2
19.2 Hudson 19.75 bo
Lafayette 32.6 5.65 17.3
Linn 17.8 .3 1.7
Lowman Hill 25.6 6.5 25.4
Lundgren 21.55 2.8 13.0
McCarter 27.4 3.0 10.9
McClure 25.1 0.0 o o
McEachron 23.5 4.3 o
Potwin 14.15 14.1 O (N
Quincy 32.35 3.1 O >-i
Quinton Heights 20.4 21.6 'T V
Randolph 27.0 ]> ró o
Shaner 21.9 os o
State Street 22.9 in lo
Stout 20.6 c?\ o
Sumner 23.3 CT\ co
Whitson 35.35 00 o
Average minority faculty/staff: 11.2%. Rec., ex. vol. IV, at 261.
. Minority Faculty/Staff In Topeka’s Secondary Schools, 1985-86
School Total Minority Per Cent
18.4 Chase 40.5 7.45
18.5 Eisenhower 65.25 12.1
5.6 French 42.6 2.4
9.6 Jardine 39.65 3.8
9.6 Landon 31.1 3.0
24.7 Robinson 49.35 12.2
Highland Park 118.25 13.65
Topeka 150.1 25.5
Topeka West 120.0 3.0
Average minority faculty/staff: 12.65%.
Rec., ex. vol. IV, at 262.
. The racially identifiable schools in 1985-86 under the _+15% standard were as follows: 1985-86
Elementary Schools (26) Min W Total % Min
61.9 Belvoir VO m n vo H Ov
57.9 Highland Park North ON o n w H VO
56.8 Lafayette N oo m O' S) n]
49.4 Quinton Heights H v© M U) h* VO
Hudson 108 124 232 46.6
Avondale East 128 162 290 44.1
Lowman Hill 155 215 370 41.9
7 schools ident. minority by enrollment
*873Elementary Schools (26) Min W Total % Min
306 Bishop W K) Oj On
261 McEachron fO W H OJ o
334 Whitson (N W ÍO W OO
240 Gage N M Ln N ca
8 schools McCarter 35 347 N ON W OO
ident. white Crestview 32 326 u\ OO W !J1
by enrollment Potwin 18 215 N W W
McClure 23 296 (N N U> h* \D
Rec., ex. supp. vol. 13, pi. ex. 155 D.
1985-86
Secondary Schools (9) Min W Total % Min
Eisen- 218 hower 230 448 48.7 1 school ident.
minority by enrollment
205 Landon 21 205 226 9.3 3 schools
Topeka ident. white
West
High 120 1391 1511 7.9 by enrollment
School
French 16 241 257 6.2
Id. pi. ex. 155 G.
When the schools in these tables are looked at in notes 57 and 58 supra, one sees that all but one of the racially identifiable minority elementary and junior high schools have minority faculty/staff percentages ranging from 17.3% to 33.3%. Six of the schools identified as racially identifiable white have minority faculty/staff percentages ranging from 0.0% to 5.6%.
. For two of these schools, Gage and Potwin, the district court specifically found that they have been predominantly white schools since the Supreme Court's decision in this case, and remain predominantly white schools adjacent to schools with higher-than-average minority student population. 671 F.Supp. at 1303.
. We do not consider this part of our opinion necessarily in conflict with the district court’s conclusion that there is no illegal segregation in Topeka, because the court did not separately consider the issue of current segregation apart from the question of causation. As we previously pointed out, the district court did find the *874existence of racial disparities in school enrollment and staff/faculty assignment. See supra at 867, 869. Our disagreement with the district court is chiefly on the significance of these findings in a school district with Topeka’s history, and bearing the weight of a presumption against it which the district court failed to accord.
. Thus, one of the school district’s experts, after reluctantly admitting that Lowman Hill elementary school's attendance boundaries were drawn in the 1950s with the effect of encompassing the only two areas of black population in that part of Topeka, and that the school was long surrounded by other schools with few to no black students, objected that "the boundaries of those school districts [were] in place.” Rec., vol. XI, at 2345.
. See Dayton II, 443 U.S. at 539 n. 11, 99 S.Ct. at 2980 n. 11 (”[F]aculty segregation disappeared completely only after efforts of the Department of Health, Education, and Welfare.”); Columbus, 443 U.S. at 461 & n. 8, 99 S.Ct. at 2948 & n. 8 (practices of assigning black teachers to schools with substantial black populations and creating optional attendance zones to permit white students to avoid predominantly black schools ceased respectively only after Ohio Civil Rights Commission complaint and the instant lawsuit); Tasby, 713 F.2d at 93 (" '[T]he District has never voluntarily moved to desegregate the Dallas school system; every step toward desegregation has been due to court action.'").
. In 1955, 3 elementary schools were 99 + % black and 14 elementary schools were 90 + % white for a total of 17 out of the 23 elementary schools. In 1966, 1 elementary school was 90 + % minority and 19 elementary schools were 90 + % white, for a total of 20 out of the 35 elementary schools. Rec., ex. vol IV, at 39-40, 54-56. Pages 30-181 of volume IV of the exhibits consist of student enrollments for individual schools from 1950 to 1985. We henceforth refer to this part of the record as Student Tables.
. In 1955, about half of Topeka’s black students attended three schools that were 99+% black. In 1966, approximately the same percentage of minority students attended one 90 + % minority school and three 50-80% minority schools. Student Tables, 1955, 1966.
. Rec., vol. XI, at 2326.
. Student Tables, 1966, 1974. Lyman elementary school was deannexed during this period.
. In 1966, Belvoir elementary school had 59.7% minority students and Lafayette elementary school had 54.5% minority students. By 1974, Belvoir had 67.1% minority students and Lafayette had 68.9%. At the secondary level, Crane junior high rose from 34.45% minority in 1966 to 52.9% in 1974, while East Topeka junior high remained more than 60% minority. Student Tables, 1966, 1974.
. Rec., vol. V, at 12-14.
. Rec., vol. XII, at 2485-91; rec., ex. vol. V, at 245-47.
. Topeka’s dissimilarity index for elementary schools dropped from 62.1 in 1955 to 51.6 in 1975, a drop of approximately 10 points. In 1981 the index stood at 40.8. Topeka’s relative exposure index for elementary schools fell from 47.7 in 1955 to 27.1 in 1975, and further to 17.2 in 1981. These drops indicate that the school system as a whole was becoming less segregated. Rec., ex. vol. II, at 156-57.
. Student Tables, 1975, 1981; compare rec., ex. vol. I, at 48-B (1974 map) with id. at 49-A (1979 map; no change to Bishop, Avondale West, or Lundgren).
. Student Tables, 1985 (McEachron and Whit-son).
. Rec., vol. XII, at 2473-75.
. Plaintiffs point to the school board’s refusal to adopt two desegregative reorganization plans proposed in 1984 by the Topeka school administration as evidence that the district is not presently committed to desegregation. These plans were violently opposed by minorities and whites alike because they would have destroyed the neighborhood school system, and we do not believe that the failure to adopt them reflects a current lack of commitment to desegregation. Nevertheless, the proposal of plans described as completely unworkable by a then-member of the Board does not satisfy the district's duty to desegregate its schools.
. Two elementary schools with more than 80% minority students were closed (Monroe 1975, Parkdale 1978) and the target minority percentages approved by HEW in 1976 have either been met or improved on. Compare rec., ex. vol. V, at 37 (projected enrollments of elementary schools under plan designed to satisfy HEW) with Student Tables, 1985. It should be noted that HEW did not approve the projected enrollments of more than 70% minority for Lafayette and Belvoir under the plan cited above. Rec., ex. vol. II, at 173. In the last five years, minority enrollments have increased in the primarily white schools, bringing them towards or over 10% minority, while the percentage of minority students at disproportionately minority schools has remained steady or dropped slightly. Student Tables, 1980-1985. The difference between schools that are more heavily minority and those that are primarily white have thus been narrowing.
. Cf. Stout v. Jefferson County Bd. of Educ., 537 F.2d 800 (5th Cir.1976) (mountain range and dangerous roads).
. Cf. Morgan v. Nucci, 831 F.2d 313, 320 (1st Cir.1987) (racially identifiable schools in heavily black areas or cut off by harbor); Ross v. Houston Indep. School Dist., 699 F.2d 218, 224 (5th Cir.1983) (segregated schools located at opposite ends of large urban school district).
. We do not mean to suggest that the school district has not faced opposition on other grounds. Both the reorganization of the late 1970s and two plans proposed in 1984 for further reorganization were met by community resistance. Rec., vol. XIII, at 2437-38.
. Cf. Tasby v. Wright, 713 F.2d 90, 93 (5th Cir.1983); Ross, 699 F.2d at 222.
. Rec., vol. XII, at 2418 (Board member 1957— 1965: no discussions on segregative or desegre-gative effects of actions); rec., vol. XII, at 2440-41 (Board member 1977-1985: Board voted for open enrollment plan although warned it might have segregative effect); rec., vol. XII, at 2491 (Board member 1973-1977: Board resisted HEW efforts because it considered system unitary then).
. Rec., vol. XII, at 2423.
. Brown II, 349 U.S. at 301, 75 S.Ct. at 757; Green, 391 U.S. at 437-38, 88 S.Ct. at 1693-94.
. According to Dr. Foster, the school district administrator in charge of personnel for 1984 stated in a deposition that he had "no knowledge of an assignment policy regarding assignment of minority personnel.” Rec., vol. V, at 632. The school district does not deny that there was no policy directed expressly at assignment. Instead, it claims that its 1963 policy against discrimination in recruitment was broad enough to cover assignment. Rec., vol. X, at 1517; rec., vol. XII, at 2428-29; rec., vol. XII, at 2692-99; rec., ex. vol. II, at 58-59; rec., ex. vol. IV, at 217-19.
. Clay closed in 1975. At the time, Clay had approximately 25% minority students while Lowman Hill had approximately 45%. Rec., vol. Ill, at 282-83. Student Tables, 1975. Polk closed in 1979, and Central Park in 1980. Rec., vol. Ill, at 290-92.
. Student Tables, 1966-1985.
. Student Tables, 1966-1985.
.Rec., vol. Ill, at 249-51, 254-55. The area assigned to Quinton Heights when Van Burén closed was apparently empty of people at the time. The school district had already extended Quinton Heights into Van Buren’s attendance area before Van Burén was closed, however. Rec., vol. Ill, at 251-53.
. Rec., vol. I, at 48.
. Rec., vol. Ill, at 291-92, rec., vol. X, at 1536, rec., ex. vol. V, at 227.
. Rec., ex. vol. I, at 48.
. The schools are Bishop, Crestview, Gage, McCarter, McClure, McEachron, Potwin, and Whitson.
. Bishop was 10.5% minority, McEachron was 10.3% minority, and Whitson was 10.2% minority-
.See, e.g., Morgan v. Nucci, 831 F.2d at 323 ("little more can be done to integrate these schools given the available numbers of white students”); Ross v. Houston Indep. Sch. Dist., 699 F.2d at 229 (district court concluded that further measures would be impractical).
. See, e.g., United States v. Lawrence County School Dist., 799 F.2d at 1047 ("greater balance of students by race achievable").
. Cf. Morgan v. Nucci, 831 F.2d 313 (1st Cir.1987) (harbour and concentration of black population in certain districts); Stout v. Jefferson County Bd. of Educ., 537 F.2d 800 (5th Cir.1976) (mountain range and dangerous roads).
. In this regard, the district court questioned defendants’ expert Dr. Armour about these possibilities. The questions and answers are instructive:
"THE COURT: ....
"Let me ask you a layman’s question here. Say we have a school that has a 55 percent minority or 60 percent minority, and we want to change that situation, and try to get it down below 50 percent minority. Has there been any experience in anything that you have run into where they have gone into what you might call a voluntary busing program where you would ask the people in that community voluntarily to come forward, and allow their children to go to another area to try to redress the racial — have you had any experience on that, or do you know of any experiences on that, or is that fanciful to think that that might happen?
THE WITNESS: Not at all. In fact that can be done. It can be done without the mandatory assignment, and has been done in many school districts, in fact with schools a lot more imbalanced than 55 percent minority.
What you try to do in those cases is twofold. One, you have an M to M program usually with transportation provided. You really encourage and do a lot of publicity and a lot of encouragement of the minority transfers out of that school, so that helps by reducing the minority population because they will frequently — minority students will frequently, given the incentives, transfer out of that school to a predominantly white school in a different part of the district.
Other [sic] thing you do in a school like that generally that is successful is you put a special program in a magnet, call them a magnet program. White parents or white students will generally not transfer out of a predominantly white area neighborhood school into a school that is predominantly minority without an academic incentive, usually because they unlike some of the minority parents they feel their school is doing a fine job. So you will need to offer an attractive program. We have learned that computer science magnets, business, health professions. Depending upon the grade level, there are programs that are very attractive to white parents, and will actually induce them to transfer into that school.
The combination of those two can frequently bring a school into the composition that you want.”
Rec., vol. XII, at 2638-39 (emphasis added).
. Rec., vol. VI, at 794-801.
. The faculty/staff analysis was performed for 1981, the only year for which there is faculty-alone data. Dr. Foster concluded that the movement of 25 faculty/staff, or approximately 3% of total faculty/staff, would satisfy plaintiffs' formula. This 25 includes the movement both out of and into schools. Rec., vol. V, at 663-64; rec., vol. VI, at 764-67, 773-74.
. Section 601 reads:
"No person in the United States shall, on the ground of race, color, or national origin, be excluded from participation in, be denied the benefits of, or be subjected to discrimination under any program or activity receiving Federal financial assistance.”
42 U.S.C. § 2000d (1982).
. The state had been dismissed as a party to the action in 1979. Brief of Appellee John Carlin at 3.
.See Milliken v. Bradley, 433 U.S. 267, 97 S.Ct. 2749, 53 L.Ed.2d 745 (1977) (affirming remedy based in part on state liability); Milliken v. Bradley, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974) (state provided funds for transporting white students but not for desegre-gative busing, enacted laws designed to delay desegregation plan and maintain segregation, rescinded city’s voluntary desegregation plan, and approved school sites and construction with segregative results); Liddell v. State of Missouri, 731 F.2d 1294, 1298 (8th Cir.) (state constitution mandated segregation until 1976 and state took no action to desegegate schools after Brown I), cert denied, 469 U.S. 816, 105 S.Ct. 82, 83 L.Ed.2d 30 (1984); Los Angeles Branch NAACP v. Los Angeles Unified School Dist., 714 F.2d 946, 948 n. 2 (9th Cir.1983) (listing state policies and acts that helped to maintain segregation), cert. denied, 467 U.S. 1209, 104 S.Ct. 2398, 81 L.Ed.2d 354 (1984). Furthermore, in both Milliken and Los Angeles Branch NAACP there was clear state law giving the state, as opposed to local school authorities, a prominant role. See Milliken, 418 U.S. at 726 n. 5, 94 S.Ct. at 3118 n. 5 (Michigan constitution assigns “the whole subject" of education to the state); Los Angeles Branch NAACP, 714 F.2d at 949 (listing California cases detailing responsibility of state organs to aid in desegregation).
. Reply Brief for Plaintiffs-Appellants, plaintiff's exhibits 45, 47.
. See, e.g., rec., ex. vol. II, at 190-235.