A world of rhetoric cannot hide the world of fact.1 No one can deny that the school systems within the Kansas City metropolitan area were racially segregated before 1954, continued to be segregated after 1954, and that virtually all remain segregated today. The clearly erroneous rule, used by the lead opinion to give blanket approval to the district court’s conclusions, was never intended to be a rule of blind deference. This is particularly true when, as here, the record does not support the district court’s conclusions. This court’s affirmance of the district court, if allowed to stand, destines the Kansas City metropolitan area to racially segregated schools and a segregated community for decades to come. The lead opinion ignores Supreme Court precedent and brushes aside thirty years of this circuit’s rulings mandating school integration.2 The lead opinion also fails to acknowledge the district court’s misunderstanding of Milliken v. Bradley, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974), and the resulting erroneous legal premises on which the district court’s conclusions are based. The effect of this pervasive error makes the lead opinion’s analysis largely irrelevant to the record before us.
It now remains for the people of metropolitan Kansas City to choose whether their community and school systems shall continue to foster an environment of racial separation. Although many might respond that freedom of choice is essential to democratic government, this proposition should be accepted only so long as equality of opportunity is not denied to those who do not have the majority will on their side. Here, the tragedy of the legalized discrimination of the past century is that it has resulted in continued denial of acceptable equal opportunities to both the black and white children in the Kansas City area. The lead opinion closes its eyes both to Missouri’s long history of legally-mandated segregation of and discrimination against *696blacks and to the uncontroverted evidence at trial that showed that for decades only the Kansas City metropolitan school district (KCMSD) provided blacks in the Kansas City metropolitan area with any educational opportunity at all. Moreover, for much of the period during which the suburban population dramatically increased, blacks were precluded by law or otherwise discouraged from residing anywhere other than the southeast corridor of the KCMSD.
Although a district court’s findings of fact may not be overturned unless clearly erroneous, Fed.R.Civ.P. 52(a), an appellate court nevertheless has the power to correct errors of law, including factual findings that are predicated on a misunderstanding of the governing rule of law. Bose Corp. v. Consumers Union of United States, Inc., 466 U.S. 485, 501, 104 S.Ct. 1949, 1959-60, 80 L.Ed.2d 502 (1984) (quoted in In re Martin, 761 F.2d 472, 475 (8th Cir. 1985)). Where, as here, the district court’s factfinding process was tainted with legal error from the start of trial by the court’s erroneous application of controlling legal principles, the district court’s findings are entitled to less than usual deference and should be reversed as clearly erroneous.3
The district court’s own conclusions and the clear weight of the uncontradicted evidence establish that constitutional violations, committed by the suburban school districts (SSDs) and HUD as well as the State of Missouri and the KCMSD, have caused significant continuing segregative effects in the SSDs. I would reverse the district court’s dismissal of the SSDs and HUD, holding that the plaintiffs have made out a prima facie case for interdistrict relief; I would remand for further evidentia-ry hearings with the SSDs and HUD reinstated as parties, with directions that new findings be made in light of this opinion and any further evidence produced. Thereafter, if the overall evidence so indicates, the district court should fashion an appropriate interdistrict remedy which incorporates the SSDs and takes into consideration the liability of HUD. The result of our divided vote is to endorse the district court’s fragmentation of the plaintiffs’ evidence and its refusal to draw obvious inferences from that evidence and to thereby uphold the district court’s denial of any form of interdistrict relief, the remedy most likely to eradicate the effects of years of legalized segregation in the Kansas City metropolitan area.
*697Several fundamental errors compel reversal of the district court’s dismissal of the SSDs and HUD. I join Judge Arnold in his conclusion that the district court erroneously interpreted Milliken v. Bradley to require that each SSD have separately committed a constitutional violation with inter-district effects before it could be included in an interdistrict remedy. Unlike Judge Arnold, I would hold that this error encompasses more than the housing violations claim but instead infects all of the district court’s findings. Rather, I would hold that the district court erred in failing to find that the SSDs committed constitutional violations with continuing significant interdis-trict effects. I would hold too that the district court erred in drawing the conclusion that the constitutional violations by the state and the KCMSD have no, or at most de minimis, continuing segregative effects on the SSDs, a finding that was falsely premised on the district court’s misapplication of Milliken and on the district court’s failure to adhere to the record as developed at trial.
The District Court’s Misconstruction of Milliken v. Bradley
First, as discussed in Judge Arnold’s opinion, the district court misconstrued Milliken when it concluded that Milliken precludes interdistrict relief in a racially segregated metropolitan area unless each school district is found to be a constitutional violator and by that violation to have caused a significant segregative effect in another district. As Judge Arnold notes, in its memorandum opinion and order dismissing the SSDs the district court prefaced discussion of the plaintiffs’ legal theories regarding SSD liability with the incorrect statement that “[t]he linchpin of an inter-district case, as declared by the Supreme Court, is whether there has been a racially discriminatory act by each defendant that substantially caused segregation in another district.”4 Jenkins v. State of Mo., No.
*69877-0420-CV-W-4, slip op. at 5-6 (W.D.Mo. June 5, 1984) (emphasis added), citing Mil-liken, 418 U.S. at 745, 94 S.Ct. at 3127. The Supreme Court’s precise language in Milliken differs significantly from this gloss by the district court and actually reads:
[I]t must first be shown that there has been a constitutional violation within one district that produces a significant segregative effect in another district. Specifically, it must be shown that racially discriminatory acts of the state or local school districts, or of a single school district have been a substantial cause of interdistrict segregation.
Milliken, 418 U.S. at 744-45, 94 S.Ct. at 3127 (emphasis added). The district court thus erected an improper proof burden for the plaintiffs to overcome. Ironically, that the district court’s interpretation of Milliken has never been the legal standard by which grants of interdistrict relief have been measured was recognized even by the defendant State of Missouri, which correctly restated the Milliken standard in its brief on appeal to this court.5
In affirming the district court’s incorrect interpretation of Milliken, the lead opinion ignores the established precedent of this circuit. In Morrilton School Dist. No. 32 v. United States, 606 F.2d 222, 228 (8th Cir.1979) (en banc), cert. denied, 444 U.S. 1071, 100 S.Ct. 1015, 62 L.Ed.2d 753 (1980), this court unanimously rejected a district court’s similarly limited interpretation of Milliken. Although the lead opinion states that Morrilton was intended to be limited to cases involving gerrymandered school district lines, so narrow a view of the acts that may constitute constitutional violations and trigger interdistrict relief is unsupportable. In Morrilton, this court found that school districts which were not themselves found to be constitutional violators nevertheless could be included in interdistrict relief where the effects of the unconstitutional actions of another (in Morrilton, the state) were felt in those school districts. Morrilton, 606 F.2d at 228. Similarly, the holding in United States v. Board of School Comm’rs., 573 F.2d 400, 410 (7th Cir.), cert. denied sub nom., 439 U.S. 824, 99 S.Ct. 93, 58 L.Ed.2d 116 (1978) (cited in Morrilton, 606 F.2d at 228-29) affirms the appropriateness of granting interdistrict relief even absent separate constitutional violations by each school district, but nowhere limits relief only to cases where the school district is found to be dependent on the state and its boundaries *699established for segregative purposes. The lead opinion distorts what our earlier opinions have held in this regard.6
Again, in Liddell v. State of Mo., 731 F.2d 1294 (8th Cir.) (en banc) (Liddell VII), cert. denied, 469 U.S. 816, 105 S.Ct. 82, 83 L.Ed.2d 30 (1984), we noted that in Liddell v. Board of Educ., 667 F.2d 643 (8th Cir. 1981) (Liddell III)
we rejected the State’s argument that the district court was without authority to formulate an interdistrict plan without finding an interdistrict violation. We also noted that voluntary interdistrict pupil exchanges “must be viewed as a valid part of the attempt to fashion a workable remedy within the City.” [Liddell III, 667 F.2d] at 651. In an order appended to that opinion, we noted that the State had been “judicially determined to be a primary constitutional violator,” and we held that an interdistrict transfer plan would be salutary and would be entirely enforceable against the State. Id. at 659. *700Liddell VII, 731 F.2d at 1303. See Little Rock School Dist. v. Pulaski County Special School Dist. No 1, 778 F.2d 404 (8th Cir.1985), cert. denied, — U.S.-, 106 S.Ct. 2926, 91 L.Ed.2d 554 (1986). Though gerrymandering is certainly one type of constitutional violation, nothing in the lead opinion explains why it is the exclusive means by which a school district may suffer significant segregative effects for which interdistrict relief may be granted.7 See Evans v. Buchanan, 416 F.Supp. 328, 340 (D.Del.1976), aff'd, 555 F.2d 373 (3d Cir.1977) (Milliken’s “specification of the deliberate drawing of lines to achieve segregation was by way of example, not limitation”).
The lead opinion’s heavy reliance on Lee v. Lee County Bd. of Educ., 639 F.2d 1243 (5th Cir.1981), and Goldsboro City Bd. of Educ. v. Wayne County Bd. of Educ., 745 F.2d 324 (4th Cir.1984), is also misplaced. Goldsboro is distinguishable on several grounds. The Goldsboro city school district had been declared unitary by a federal district court in 1973. See id. at 325-26. Moreover, in Goldsboro, there was no showing of any housing violations or inter-district transfers for segregated purposes, and on appeal the Fourth Circuit specifically found that the district court did not erroneously apply the law to that case’s facts. See id. at 328. Although Lee exhibits some similarities to the present case, as we noted in Little Rock the Lee case did not involve segregative interdistrict transfers, segregative boundary changes, or state-imposed residential segregation, and the city district had been previously declared unitary. See Little Rock, 778 F.2d at 429.
Missouri’s History of De Jure Segregation
Significantly, the lead opinion omits any discussion of the State of Missouri’s long history of de jure segregation in which the KCMSD and most, if not all, of the SSDs participated. Much of this history, which was recounted in Liddell v. Board of Educ., 731 F.2d 1294 (8th Cir.1984) (Liddell VII), is detailed in the district court’s opinion granting intradistrict relief. See Jenkins v. State of Mo., 593 F.Supp. 1485, 1490-95 (W.D.Mo.1984). This history bears repeating here, however, because the nature and gravity of the constitutional violations for which plaintiffs seek redress can properly be evaluated only in light of their historical context.
The State of Missouri’s implementation and enforcement of school segregation can be traced to well before the Civil War, when Missouri enacted laws prohibiting the creation of schools to teach blacks to read or write. Act of February 16, 1847, § 1, 1847 Mo.Laws 103. Beginning in 1865, the Missouri constitution expressly required separate schools for blacks and whites. Mo. Const. 1865, art. IX, § 2. This provision was retained in three successive constitutions. Mo. Const. 1875, art. XI, § 3, retained Mo. Const. 1919, art. XI, § 3, *701revised and retained, Mo. Const. 1945, art. IX, §§ 1(a) and 3(c). See also Act of February 17, 1865, § 13, 1865 Mo.Laws 170. In 1889, the Missouri legislature made it a criminal offense for a black child to attend a white school. Act of June 11, 1889, § 7051(a), 1889 Mo.Laws 226. Although a 1954 Attorney General opinion declared the school segregation laws unenforceable, the statutes implementing the constitutionally mandated segregation were not repealed until 1957, see Act of July 6, 1957, § 1, 1957 Mo.Laws 452, and the constitutional provision was not rescinded until 1976. See Adams v. United States, 620 F.2d 1277, 1280 (8th Cir.), cert. denied, 449 U.S. 826, 101 S.Ct. 88, 66 L.Ed.2d 29 (1980).
Prior to 1954, school districts in Missouri were not even in compliance with the “separate but equal” doctrine of Plessy v. Ferguson, 163 U.S. 537,16 S.Ct. 1138, 41 L.Ed. 256 (1896). From 1866 to 1929, state law exempted school districts from providing schools for black children if there were fewer than fifteen black children in the district and required them to discontinue black schools whenever average daily attendance fell below eight students. See 1865 Mo.Laws 177; 1869 Mo.Laws 86-87; 1870 Mo.Laws 149; 1887 Mo.Laws 264; 1893 Mo.Laws 247; 1909 Mo.Laws 790-91. In 1929, the Missouri legislature gave school districts the option to forgo schools for blacks altogether, no matter how many black school children resided in the district. 1929 Mo.Laws 382. Although school districts refusing to provide schools for blacks within their district were required to make interdistrict arrangements for and pay the tuition of black elementary school children beginning in 1929 and for black high school students beginning in 1931, the school districts were not required to fully reimburse transferred black students for transportation costs until 1945. See 1945 Mo.Laws 1700.
Rather than providing schools for blacks within their districts, school districts in the Kansas City metropolitan area, including the predecessor districts of the SSDs,8 took full advantage of state law provisions allowing them to arrange for the transfer of black school children out of their districts. Blue Springs, Fort Osage, Center, Ray-town, Grandview, and Hickman Mills provided no schools for blacks.9 Lee’s Summit did not operate any black schools after 1910. North Kansas City intermittently operated a black elementary school until 1944. Independence, Liberty, and Park Hill each operated a black elementary school. Independence operated a black high school from approximately 1934 to 1945, and Liberty offered high school through the eleventh grade after 1938. However, the evidence is overwhelming that the quality of education available in these black schools was quite poor, especially in comparison to that available in the white schools in the SSDs and the black *702schools in KCMSD. Moreover, despite state laws requiring districts not operating black schools to reimburse black students for tuition and at least part of their transportation costs, none of the SSDs did so until Lee’s Summit began reimbursement in 1931, followed by Independence in 1945, Park Hill and North Kansas City in the late 1940s, and Liberty in 1953. As the district court itself found, in 1954 KCMSD was the only school district in the Kansas City metropolitan area providing any real educational opportunities for blacks, operating fourteen black elementary schools, one black junior high vocational school and one black high school-junior college. See Jenkins, 593 F.Supp. at 1492.
Legalized segregation of and discrimination against blacks in Missouri was not limited to education. State law permitted local authorities to establish separate libraries, parks, and playgrounds for whites and blacks, Mo.Rev.Stat. § 165.377 (1959); made it a crime for a black to marry a white, Mo.Rev.Stat. § 563.240 (1959); and required segregation in colonies for the “feebleminded.” See, e.g., Jenkins, 593 F.Supp. at 1503. Of foremost importance to this case is the pervasive system of housing segregation in which all levels of government participated. Racially restrictive covenants, FHA appraisal practices, relocation policies of state and local housing agencies, and private discrimination operated to restrict blacks to certain geographic sections of the Kansas City metropolitan area.
Housing Violations and HUD Liability
Judge Arnold’s opinion recounts the long history of Missouri’s involvement with housing discrimination in the Kansas City area, which in some instances continued through the 1970s. I join Judge Arnold in concluding that remand is required for further findings, for the purposes of fashioning interdistrict relief, on the extent of constitutional violations by the state and its various agencies in implementing segre-gative housing policies. For similar reasons, I would also reverse and remand for further consideration the district court’s dismissal of HUD. Though the lead opinion affirms the district court’s finding that HUD did not act arbitrarily or capriciously and that HUD’s acts were without discriminatory intent or effect, the record makes clear that HUD committed numerous constitutional violations. As Judge Arnold points out:
With regard to HUD, the District Court found that the agency had acted “reasonably” because it had taken action in the 1970s to end discriminatory practices by HAKC, whose programs HUD funded. 593 F.Supp. 1498-99. Yet there was extensive evidence, not addressed by the District Court, that HUD knew of shortcomings in HAKC’s tenant-assignment practices for years without taking action to end them.
Ante at 692 n. 10. Use of the arbitrary and capricious standard is inappropriate in the context of HUD’s alleged constitutional violations through the participation in and support of racially segregated housing practices. Where a government entity’s discriminatory practices causes housing and population shifts with metropolitan-wide effects, an interdistrict remedy has been found an appropriate form of relief. United States v. Board of School Comm’rs., 637 F.2d 1101, 1114 (7th Cir.), cert. denied, 449 U.S. 838, 101 S.Ct. 114, 66 L.Ed.2d 45 (1980). Moreover, considering whether the participation of each SSD in the dual school system had any significant interdistrict effect, the district court concluded that for each SSD the proof of any current effects was “weak, speculative and in any event de minimus [sic].” Jenkins, June 5 opinion at 12. Because the district court erroneously assumed that each SSD must be found to have committed a constitutional violation with current segregative effects before it could be included in inter-district relief, the district court did not properly take into account the current segregative effects on the SSDs of the state’s or KCMSD’s actions to maintain a dual school system in the Kansas City metropolitan area. *703Milliken makes clear that an interdistrict remedy is justified if the constitutional violations of even “a single school district have been a substantial cause of interdis-trict segregation.” Milliken, 418 U.S. at 745, 94 S.Ct. at 3127. The Supreme Court denied interdistrict relief in Milliken simply because the plaintiffs had only shown that the constitutional violations of the Detroit public school system had mtradistrict, not ráíerdistrict, effects. Id. at 747, 94 S.Ct. at 3128 (explaining that dissent’s position was that the existence of a dual school system in Detroit justified an interdistrict remedy); id. at 749, 94 S.Ct. at 3129 (“Where the schools of only one district have been affected, there is no constitutional power in the courts to decree relief balancing the racial composition of that district’s schools with those of surrounding districts.”); id. at 751, 94 S.Ct. at 3130 (“Thus, there was no evidence suggesting that the State’s activities * * * within Detroit affected the racial composition of the school population outside Detroit or, conversely, that the State’s * * * activities within the outlying districts affected the racial composition of the schools within Detroit.”). In contrast, the plaintiffs here presented evidence showing that KCMSD’s decision to operate some completely black schools and some completely white schools long after 1954 led to the rapid turnover of neighborhoods from black to white and caused many whites to move out of KCMSD to the SSDs. The district court repeatedly recognized that evidence at trial demonstrated the link between KCMSD’s delay in implementing effective integration policies during the 1960s and 1970s and the exodus of whites from KCMSD during this period in discussing KCMSD’s liability for segregation within KCMSD. See Jenkins, 593 F.Supp. at 1492-95. The district court did not consider KCMSD’s conduct as a basis for ordering interdistrict relief, however, because it misconstrued Milliken to require culpability with continuing segre-gative effects on the part of each SSD before that SSD could be included in an interdistrict remedy. Because under Mil-liken the conduct of even a single school district may be the basis for interdistrict relief if the continuing effects of that conduct are interdistrict, I would require the district court to consider on remand the extent to which KCMSD’s delay in implementing an effective integration policy has current interdistrict effects.
Ironically, though Judge Arnold concludes that the district court misapplied Milliken in assessing interdistrict effects from housing violations, he apparently upholds as not clearly erroneous the district court’s findings of no constitutional violations by the SSDs and its findings of only de minimis continuing segregative effects on the SSDs from the state’s and the KCMSD’s operation of a dual school system. To this extent, I part company with Judge Arnold.10 These findings are inseparable from the district court’s misunderstanding of the Milliken standard.
Constitutional Violations by the SSDs
The district court failed to analyze the housing violations by the state agencies because it believed they were irrelevant to the SSDs’ participation in any interdistrict relief. Similarly, the district court failed to analyze the effect of the state’s and the KCMSD’s discriminatory practices because it decided their conduct was irrelevant to any remedial relief. The record is replete with evidence of constitutional violations by many of the SSDs and of the effect those violations had on all of the SSDs and on the KCMSD. As has already been stated here, these violations cannot be ignored by simply expressing deference to the district court’s findings under the clearly erroneous rule. Whether there exists a constitutional violation is more a question of law than of fact and is clearly subject to de novo appellate review. See, e.g., Bose *704Corp., 466 U.S. at 501 n. 17, 104 S.Ct. at 1960 n. 17 (“A finding of fact in some cases is inseparable from the principles through which it was deduced.”).
Even under the district court's erroneous interpretation of Milliken, the lead opinion errs in affirming the district court's dismissal of the SSDs because it failed to find that the SSDs committed constitutional violations with cross-district effects. The un-controverted evidence at trial showed that under the pre-1954 dual school system the SSDs exercised their discretion, granted them by the state, either to provide schools for black students which offered at best a substandard education or to decline, to provide any schools at all. Instead, the SSDs transferred black students to the KCMSD. Understandably and undisputedly, KCMSD became the only school district in the Kansas City metropolitan area in which black school children could receive a complete education. Not only did the SSDs thus create segregated conditions in their individual districts by emptying their districts of all black school children, but they set historical precedent for interdistrict transfer of students on the basis of race in the Kansas City metropolitan area.
The district court’s conclusion that the SSDs were local automonous units11 and therefore exempt from liability for any seg-regative acts done by the state is the major premise on which the lead opinion relies in affirming the district court’s finding that the SSDs committed no acts with discriminatory intent that had significant segre-gative effects. However, to the extent the SSDs exhibited autonomy in deciding to provide inferior or no schools for blacks, under Milliken proof of autonomy here leads only to the conclusion that the SSDs did commit constitutional violations. Moreover, the lead opinion ignores the uncontra-dieted fact that after the SSDs participated in pre-1954 segregation on an interdistrict level, they took no affirmative steps on a similar interdistrict level to reverse the effects of those autonomously chosen educational practices. Even if Milliken required proof of constitutional violations by each individual SSD, which it does not, the district court erred in failing to draw the obvious inferences from the uncontrovert-ed evidence to find such violations in the “autonomous” acts of the SSDs. The result today, as the evidence at trial showed, is that KCMSD’s student body is sixty-eight percent black whereas the student population of the SSDs is only five percent black.12
Continuing Significant Interdistrict Effects
My most fundamental objection to the majority’s position, however, is its wholesale affirmance of the district court’s finding that to the extent any constitutional violations were committed by any of the defendants, the continuing segregative effects on the SSDs were de minimis and that inclusion of the SSDs in an interdistrict remedy was therefore precluded.
In concluding that the pre-1954 dual school system in which the SSDs participated has no continuing interdistrict effects, the district court found that despite the fact that only KCMSD provided blacks in the Kansas City metropolitan area with *705any real educational opportunities prior to 1954 as a direct result of the constitutional violations of the state, the KCMSD and the SSDs, those violations had but a de minimis effect on the housing choices of blacks prior to 1954. Such a finding is contrary to legal precedent, many of the district court’s own factual findings, and the plaintiffs’ unrebutted expert and lay witness testimony and is clearly erroneous.
In affirming as not clearly erroneous the district court’s findings of de minimis continuing segregative effects on the SSDs from the KCMSD’s actions, the lead opinion follows the district court in overlooking the weight of uncontradicted evidence in the record and the district court’s own findings that the KCMSD’s perpetuation of segregated schools within the KCMSD caused a mass concentration of blacks in an area where a substantial number of whites lived. This in turn resulted in displacement of whites, who moved to or transferred into the SSDs.
The evidence produced at trial demonstrated that by a variety of affirmative official acts, the KCMSD sought to preserve the all-white character of some of its schools at the expense of the schools in the southeast corridor of Kansas City. There, KCMSD followed a policy of “integration and stabilization.” As part of this piecemeal desegregation policy, KCMSD implemented attendance zones, intact busing, liberal transfers, neighborhood school policies, and faculty transfers which had segre-gative instead of integrative effects. One of the most significant of these was the creation of shifting and optional attendance zones. Evidence at trial indicated that the attendance zone choices were based on Troost, a north/south street that traditionally had divided white neighborhoods west of Troost from the neighborhoods to the east which also contained many whites but in which most of the blacks in Kansas City lived. Testimony at trial indicated that despite overcrowding in schools east of Troost and underuse of schools west of Troost, attendance zones did not traverse Troost until 1976. Moreover, testimony at trial established and the district court itself found that as the population of Kansas City changed in the years following Brown v. Board of Educ., 349 U.S. 294, 75 S.Ct. 753, 99 L.Ed. 1083 (1955), KCMSD “made frequent shifts in the attendance areas of its schools” with the result of removing white sections from the western-most portions of the racially transitional areas and attaching them to all-white zones further west, shifts which, coupled with a liberal transfer policy, “allowed attendance patterns to continue on a segregated basis.” Jenkins, 593 F.Supp. at 1494. The district court also found that “as the black population expanded from the central city in a southeast direction, neighborhoods and schools experienced a racial transition. As blacks moved, or were bused to the schools in the [southeast] area, whites moved out.” Id.13 (citations omitted).
As a result of these policies, blacks were bused to or moved to the southeast area of Kansas City and whites moved to the neighboring SSDs. As the black population of Kansas City expanded to the southeast, KCMSD chose to operate some completely segregated schools and some integrated ones. The result of KCMSD’s official school policies was to make the KCMSD “blacker” and the SSDs “whiter,” actions which have been found by courts before to be constitutional violations for which interdistrict relief is appropriate. Cf. Hoots v. Pennsylvania, 672 F.2d 1107, 1121 (3d Cir.), cert. denied, 459 U.S. 824, 103 S.Ct. 55, 75 L.Ed.2d 60 (1982) (quoting Penick v. Columbus Bd. of Educ., 429 F.Supp. 229, 266 (S.D.Ohio 1977), aff'd, 443 U.S. 449, 99 S.Ct. 2941, 61 L.Ed.2d 666 (1979)) (“Actions and omissions by public *706officials which tend to make black schools blacker necessarily have the reciprocal effect of making white schools whiter.”) (citation omitted). See Evans v. Buchanan, 393 F.Supp. 428, 436 and n. 15 (D.Del.1975) (presence of proportionately more black children in a city school than in the neighborhood served by the school likely encouraged white families to move to nearby suburbs, where “black students were barely present” in the school systems).
It is evident that the segregative policies perpetuated by KCMSD caused a greater concentration of blacks and black schools in the southeast corridor, directly resulting in displacing whites into the SSDs.14 In this regard, the district court’s own findings are at least contradictory and at most support the grant of interdistrict relief based on continuing effects in the SSDs from the KCMSD’s policies. The district court recognized that the KCMSD had chosen to operate some completely segregated schools and some integrated ones in the district and that as of 1977, twenty-five one-race schools under the pre-1954 dual system remained ninety percent or more of the same race. See Jenkins, 593 F.Supp. at 1492. Moreover, in a supplemental order, of which we take judicial notice, that enjoined for one year the property tax rollback provided under a Missouri statute, the district court summarized its findings in this case to date by stating:
The Court has found that segregated schools, a constitutional violation, has led to white flight from the KCMSD to suburban districts, large number of students leaving the schools of Kansas City and attending private schools and that it has caused a system wide reduction in student achievement in the schools of KCMSD. It has also found that by improving the quality of education in the KCMSD, it will enhance the appeal of the school system, thus giving it a chance to retain its present white enrollment and also encourage whites in private and suburban schools to enroll in its schools. As set forth in its order of June 14,1985, the basic remedial principle in school desegregation cases, is that “the scope of the remedy is determined by the nature and extent of the constitutional violation.” Milliken v. Bradley, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974).
Jenkins v. State of Mo., No. 77-0420-CV-W-4, slip. op. at 1-2 (W.D.Mo. August 25, 1986). Not only does this language further support the view that the district court did recognize that the effects on the SSDs from the segregative policies of the KCMSD were more than de minimis, but it illustrates that the district court’s refusal to grant interdistrict relief was clearly erroneous and contradictory to its own findings from the record.15
That the availability of schools influences housing choices is a common sense proposition. After a fifty-year period during which the SSDs provided no schools or inadequate schools for blacks, it is hardly surprising that black families living in the SSDs and those moving to the Kansas City area from other states chose to move to the KCMSD, where adequate educational facilities were available. As the district court notes in its opinion on intradistrict liability, the nexus between availability of schools and housing patterns has been recognized by the Supreme Court. See Jenkins, 593 F.Supp. at 1491 (quoting Swann, 402 U.S. *707at 20-21, 91 S.Ct. at 1278-79). This nexus has also been recognized by this court in the context of interdistrict cases. See, e.g., Little Rock, 778 F.2d at 412 n. 5. The Little Rock case is particularly instructive in that, under facts very similar to this case, we upheld a district court finding that black families from other school districts were drawn to the Little Rock school district precisely because it was the only district that provided blacks in the area with substantial educational opportunities. Id. at 412. The district court’s finding in Little Rock that school availability influences housing on an interdistrict as well as an intradistrict basis, which we upheld, seriously undermines the credibility of the district court’s finding in this case “that the absence of black schools in any of the [SSDs] did not discourage black families outside (or from within) Missouri from moving to and living in those districts.” Jenkins, June 5 opinion at 18 (citations omitted).
The district court's finding is further undermined by several statements in its September, 1984 opinion regarding intradistrict liability of the state and KCMSD. In rejecting the state’s argument that job availability rather than school availability was responsible for the segregated housing and school patterns within the KCMSD, the district court stated that “[ojften jobs would pull migrants to the city and then availability of schools would influence, more specifically, what housing choice would be made within the city.” Jenkins, 593 F.Supp. at 1490. In finding that the current dual housing market affected blacks in the KCMSD, the court again referred to the “inextricable connection” between schools and housing. Jenkins, 593 F.Supp. at 1491. It is difficult to see how school availability would be an important factor in making housing choices within a single school district, but would not influence housing decisions within the metropolitan area in general.
In support of their argument that the constitutional violations here produced significant interdistrict effects, plaintiffs produced statistics showing that although the school population in the three-county area, excluding KCMSD, was between 7.19% and 8.7% black in 1881, the black school-aged population in the same area had steadily declined to between .39% and 1.5% by 1954. At the same time, the black population in KCMSD had risen from 8.8% in 1900 to 14.0% in 1954. Plaintiffs produced several expert witnesses on education and housing patterns who testified that the inability of blacks to provide their children with an education in the SSDs caused many black families to move to KCMSD, the only district providing any real educational opportunities for blacks in the Kansas City metropolitan area. The absence of schools for blacks in the SSDs, the experts testified, also influenced the housing choices of blacks moving into the Kansas City metropolitan area from other states, both because there would be less chain migration to the SSDs as the existing black population in the SSDs declined and because accessible education is an important factor for blacks, as for all people, in making housing choices. Plaintiffs also produced the testimony of many blacks who had lived in the SSDs prior to 1954 but had had to travel long distances or move to KCMSD itself to receive an education. Other evidence included pre-1954 school board reports and other documents stating that the absence of educational opportunities for blacks in rural areas caused by the dual school system was causing many black families to leave rural areas for large cities.16
Notwithstanding plaintiffs’ extensive evidence, the district court concluded that the absence of educational opportunity for blacks in the SSDs prior to 1954 had a de minimis effect on current housing patterns and school segregation in the Kansas City metropolitan area. The lead opinion merely reasserts the district court’s findings and concludes that they are not clearly erroneous. Close examination of the district court’s opinion, however, reveals that *708its reasons for rejecting the plaintiffs’ evidence as de minimis ring hollow. For example, the district court found that the increase in the black population in KCMSD prior to 1954 was attributable to jobs and employment opportunities available in the KCMSD area, rather than to the availability of schools. Jenkins, June 5 opinion at 15-16 and 18. No one disputes that economic opportunity was a major factor in drawing blacks, especially from the South, to the Kansas City metropolitan area. The plaintiffs’ argument, however, was not that school availability to the exclusion of economic opportunity drew blacks to KCMSD. Rather, plaintiffs argued that once blacks were drawn to the Kansas City area for whatever factors, including economics, school availability caused many blacks to settle in KCMSD rather than in the SSDs. This argument remains unrefuted, and in fact is substantiated by the district court’s own findings regarding the nexus between school availability and housing patterns.17 See Jenkins, 593 F.Supp. at 1491.
The district court also failed to assess the importance of the relative depopulation of blacks from the SSDs prior to 1954 with respect to the present racial composition of the Kansas City area. In rejecting the plaintiffs’ evidence of depopulation, the district court noted that although the black population in the three-county area, excluding KCMSD, decreased by only approximately 837 from 1910 to 1960, the black population in KCMSD increased by more than 45,000 during this same period. Thus, the district court concluded, the impact of that movement on the black enumeration in KCMSD was insignificant. Jenkins, June 5 opinion at 15-16. The effect of the decline of the black population living in the SSDs prior to 1954, however, is not limited to those black families that left the SSDs for KCMSD because of the lack of schools. Instead, as the plaintiffs’ evidence demonstrated, the resulting lack of blacks in the SSDs in turn caused blacks moving to the Kansas City area from the South to live in the KCMSD, where there was a growing black population, rather than in the overwhelmingly white and hostile SSDs.
Not only did the district court improperly discredit the plaintiffs’ evidence, it also omitted any reference to the evidence that most strongly supported plaintiffs’ claims. For example, the plaintiffs submitted several reports prepared by the superintendent of public schools in Missouri during the 1920s and 1930s stating that the absence of black schools in rural areas was causing black families to migrate to the large cities.18 Contemporaneous documents such as these, prepared by school officials at the time black depopulation of rural areas was occurring in Missouri, are highly probative of the link between school availability and housing patterns on an interdistrict basis. Although the district court recognized this link, see Jenkins, 593 F.Supp. at 1490 (“[ujndeniably, some blacks moved to districts, including the KCMSD, that provided black schools”), it largely discounted the plain inferences to be drawn from it.
The district court rejected the plaintiffs’ statistical evidence regarding the number *709of black students in the SSDs who trans-fered to KCMSD prior to 1954 as “de min-imis.” It is true that the number of transfers actually shown to have occurred during this period, anywhere from 251 to 600, appears small. These examples, however, were presented only as representative of the effects of the dual school system on school attendance and housing patterns in the Kansas City area, not as an exhaustive enumeration. The plaintiffs’ inability to produce more accurate statistical evidence of interdistrict transfers is not surprising in light of the incentives to underreport black students and the conceded unavailability of interdistrict transfer records, for which the SSDs are at least partially responsible. The SSDs had an incentive to underreport the number of black school children in their districts because state law excused them from their duty to provide a school for blacks if the number of black school children in the district fell below a certain number. Black students from the SSDs had an incentive to be counted as residents of the KCMSD rather than as residents of the SSDs because they could thereby avoid paying tuition to attend KCMSD before the SSDs began to reimburse them for tuition in the 1930s and 1940s. It is difficult to imagine what level of statistical proof would satisfy the district court and the lead opinion. Evidently, they would be satisfied only if the plaintiffs had paraded before the district court every black student who had transfered from the SSDs to KCMSD or who had moved from the SSDs to KCMSD prior to 1954. Such a stringent level of proof has never been required. See, e.g., Board of School Comm’rs., 637 F.2d at 1113-14.
Even assuming the numbers relied upon by the district court are accurate and controlling, it requires little additional evidence to buttress the common sense proposition that blacks would be deterred from migrating to an overwhelmingly white school district where no black schools existed, where arbitrary policies or school transfers placed the primary burden of educational expenses on many black families, and where racial covenants prevented home purchases. To fail to acknowledge that blacks were deterred from living in the SSDs is to ignore the obvious. Cf. International Brotherhood of Teamsters v. United States, 431 U.S. 324, 367, 97 S.Ct. 1843, 1870-71, 52 L.Ed.2d 396 (1977) (in reversing denial of Title VII relief, the Supreme Court stated that to fail to account for non-applicants being deterred from even applying for employment would mean that “[v]ictims of gross and pervasive discrimination could be denied relief precisely because the unlawful practices had been so successful”). Perhaps most disturbing, however, is the district court’s repeated reliance on the increasing black enrollment in the SSDs after 1954 as an indication that the prior constitutional violations of the SSDs have no current inter-district segregative effects. The district court noted, for example, that black enrollment in Hickman Mills increased from twenty-nine black students in 1967 to 1,571 in 1983, and that black enrollment in Park Hill has increased by 110 students since 1976. Jenkins, June 5 opinion at 56, 88. That black enrollment in the SSDs has increased since the 1960s may indicate that barriers to black movement into the SSDs are declining over time. Far from indicating that there are no current significant segregative effects of the SSDs’ past conduct, however, this increase in black enrollment in the SSDs since Brown v. Board of Educ. is conclusive evidence that the lack of schools for blacks in the SSDs prior to 1954 was a deterrent to blacks settling there. In light of the compelling evidence presented by the plaintiffs that the absence of schools for blacks in the SSDs prior to 1954 had a significant effect on housing choices made by blacks during this time, the district court’s conclusion that the effects were de minimis is clearly not supported by the record.
Affirmative Duty of State of Missouri, KCMSD, and the SSDs
It is clear that the SSDs, as well as the State of Missouri and KCMSD, have all failed to fulfill their affirmative duty under Brown v. Board of Educ. to eliminate the *710vestiges of their prior segregative policies and that significant interdistrict effects of these policies are still evident in the Kansas City metropolitan area. The district court, now affirmed by this court, held that because all of the SSDs had opened the doors of their all-white schools to blacks in 1960, the SSDs had discharged their duty under Brown. Courts have long recognized, however, that racially neutral policies are generally ineffective to counteract the continuing effects of past segregative practices, especially where such segregation has influenced housing patterns. See, e.g., Swann, 402 U.S. at 20-21, 91 S.Ct. at 1278-79; Kemp v. Beasley, 389 F.2d 178, 190 (8th Cir.1968). The district court itself recognized the ineffectiveness of racially neutral policies after a long period of intentionally segregative policies in its opinion on intradistrict liability. See Jenkins, 593 F.Supp. at 1490-95. The reason such racially-neutral plans are ineffective is plain. A racially neutral policy, such as a neighborhood school policy, will simply not correct the effects of many years during which black schools were located in only certain parts of a school district, causing black families to settle in those areas. Similarly, to merely open to blacks the doors of schools in virtually all-white districts simply fails to eradicate the effects of many years when those districts provided blacks with no schools or at best inadequate schools, causing black families already living in those districts to leave and discouraging black families newly moving into the area from settling in those districts. It is precisely because of this continuing affirmative duty under Brown that pre-1954 evidence is relevant.
Conclusion
The effects of the SSDs’ pre-1954 constitutional violations and their continuing failure to fulfill their duty under Brown axe still evident in the Kansas City metropolitan area today. Eighty-seven percent of the black students in the Kansas City metropolitan area are educated in KCMSD; eighty-nine percent of the white students attend school in one of the SSDs. Yet the district court concluded that the pre-1954 conduct of segregated schools and housing practiced by the SSDs did not significantly contribute to the current racial imbalance in the Kansas City metropolitan area. To support this conclusion, the district court noted that KCMSD was only 18.9 percent black in 1954-55. The fact that the major increase in the black population in KCMSD occurred after 1954, said the district court, is strong evidence that the current racial composition of the Kansas City area was caused by factors other than the pre-1954 dual school system. See, e.g., Jenkins, June 5 opinion at 41-42. Undeniably, many factors have contributed to the enormously disparate racial composition of the student bodies of KCMSD and the SSDs. Housing policies of federal, state, and local agencies have contributed to the dual housing market that exists in the Kansas City metropolitan area, which in turn has contributed to the de facto dual school system that exists today. Ineffective integration policies of KCMSD during the 1960s and 1970s caused an exodus of whites from KCMSD to the SSDs.19 The evidence at trial showed that the continuing influx of blacks to the Kansas City metropolitan area from the South and the higher birthrate among blacks have also contributed to the higher percentage of blacks in KCMSD. To the extent that these factors have magnified the effects on the SSDs of the pre-1954 conduct of state, the KCMSD, and the SSDs and those effects have not been affirmatively addressed, all the defendants are at least partially to blame. Had the SSDs not contributed to the depopulation of blacks from those areas by failing to provide schools for blacks, the higher birthrate among blacks would have presumably also increased the percentage of blacks living in *711the SSDS. Similarly, had the SSDs’ policies not contributed to the depopulation of blacks from those areas, more blacks moving into the Kansas City metropolitan area from the South would have settled in the SSDs rather than in KCMSD.
The lead opinion agrees with the state and the SSDs, who emphasize that the plaintiffs’ case rests entirely on racial disparities, that proof of disparate impact on the SSDs is not enough to hold the SSDs liable and provide the basis of interdistrict relief. See Columbus Bd. ofEduc. v. Penick, 443 U.S. 449, 99 S.Ct. 2941, 61 L.Ed.2d 666 (1979). I have no quarrel with that premise. But as Peniek recognizes, in Swann intradistrict relief was premised on a presumption of continuing racial imbalance within a unitary district. See Penick, 443 U.S. at 479, 99 S.Ct. at 2978. Here the racial imbalance is clear evidence of continuing segregation following a pre-1954 de jure policy. It is not necessary to switch the burden of proof where there is overwhelming evidence of violations and effects throughout the system.
Moreover, the lead opinion is also incorrect in affirming the district court’s reliance on the mere passage of time as exculpating the SSDs. The lapse of decades, without any showing of affirmative steps taken to reverse the invidious effects of de jure segregation, reinforces the broad scope of the effects of the defendants’ constitutional violations here. Nor is the lead opinion correct in affirming the district court’s rejection of the overall thrust of plaintiff’s proof of interdistrict effects on the grounds that a series of de minimis violations cannot be aggregated to produce a finding of a constitutional violation. See, e.g., Dayton Bd. of Educ. v. Brinkman, 433 U.S. 406, 413, 97 S.Ct. 2766, 2772, 53 L.Ed.2d 851 (1977). Here, neither the violations found proved by the district court nor the effects of those violations are insignificant. See, e.g., Jenkins, 593 F.Supp. at 1503 (finding that presence of dual schools had important segregative effects throughout the Kansas City metropolitan area). The premises on which the majority bases its affirmance of the district court’s conclusion of de minimis continuing segregative effects on the SSDs have no basis in law nor in the facts as otherwise found by the district court and as developed at trial.
Although the lead opinion suggests to the contrary, this case differs notably from Milliken and this circuit’s Little Rock case in that it does not present on appeal an issue of whether the only appropriate remedy is consolidation of the entire metropolitan area’s school districts.20 The plaintiffs and the KCMSD seek alternative forms of interdistrict relief as well as consolidation as a means to achieve an integrated school system. And even if the requested relief includes consolidation, relief can be tailored by the district court itself or as directed by this court to fit the constitutional violations. As was true in the Little Rock case, the automony of the city and suburban school districts can be preserved concurrent with mandated interdistrict relief. As we stated in Liddell VII:
[R]elating the remedy to the violation pursuant to Milliken II [433 U.S. 267, 97 S.Ct. 2749, 53 L.Ed.2d 745 (1977)] does not require a finding that each educational program at issue has in the past been “infected with the discriminatory bias of a segregated school system.” * * * It is sufficient to determine that the remedial program is directed to cure the general condition offending the Constitution.
Liddell VII, 731 F.2d at 1315-16 n. 18 (citation omitted). The relief sought here is narrowly tailored to the particular situation and responsive to the Supreme Court’s concern in Milliken that the “nature of the violation determine[s] the scope of the remedy.” Milliken, 418 U.S. at 738, 94 S.Ct. at 3124 (citation omitted).
*712The quarrel I have with the intradistrict relief mandated by the district court, as modified by the lead opinion, is that the intradistrict programs will do little to integrate the metropolitan school system. The lead opinion calls for improved facilities and quality education programs, requires the state to pay its share of the cost of those programs, and authorizes the district court to order a tax increase if that alternative is necessary to enable the school district to bear its share of the cost of the mandated programs. This remedy is fine as far as it goes; however, it falls far short of the relief that should be required under the facts of this case.21 It is hoped that at the very least the magnet school component and the voluntary interdistrict transfer plan, if implemented, will be of some help in integrating the district. In my view, however, these interdistrict programs should have been mandated already.
To summarize, I would hold that the district court erred in interpreting Milliken v. Bradley, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974), to preclude interdis-trict relief. Under Milliken, properly interpreted, the district court’s own factual findings with regard to the constitutional violations of the state and KCMSD, especially with regard to the significant link between housing patterns and school availability in the Kansas City metropolitan area, strongly suggest that the plaintiffs are entitled to an interdistrict remedy. Because the district court clearly erred in finding that the SSDs committed no constitutional violations with continuing significant interdistrict effects, the district court’s dismissal of the SSDs should be reversed and remanded for further proceedings. If on remand, after each of the SSDs have had an opportunity to present evidence, it remains unrebutted that the acts of the state, the KCMSD, or any of the SSDs have a continuing interdistrict effect, those SSDs which the district court determines to be within the violations’ effects should be ordered to participate in an inter-district remedy, such as a mandatory inter-district transfer program, narrowly tailored to remedy the constitutional violations proved here.
The case should be reversed and remanded for reconsideration in light of Milliken v. Bradley, and the SSDs and HUD should be joined to determine the extent of their liability.
. Cf. T.H. Huxley, Lay Sermons, Addresses, and Reviews (1891) ("A world of fact lies outside and beyond the world of words").
. For the history of this circuit’s school desegregation rulings, see Heaney, Busing, Timetables, Goals, and Ratios: Touchstones of Equal Opportunity, 69 Minn.L.Rev. 735 (1985).
. The lead opinion’s reliance on Anderson v. City of Bessemer City, 470 U.S. 564, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985), provides the basis of its criticism that this dissent ”engag[es] in an original evaluation of the voluminous and ofttimes contradictory record seeking an ultimate conclusion more satisfactory than that reached by the district court." Ante at 668. It suggests that this dissent “embarks upon fact finding of [its] own" and makes "no effort to consider [the district court’s finding] in a proper Rule 52(a) analysis." Ante at 677 n. 25. It also urges that the dissent "completely and totally ignores the district court’s findings that any pre-1954 acts had negligible and de minimis current effects.” Ante at 680 n. 28.
I start with the legal premise, never refuted by the lead opinion, that the district court's opinion is fundamentally unsound because it is based on an erroneous legal premise in its interpretation of Milliken and because it totally ignores the historical and factual record. Not only is the record contradictory, as the lead opinion admits, but in its own opinions the district court contradicts itself. This dissent does not substitute the views of appellate judges for that of the trial court. What this dissent focuses on is that the district court’s application of the law to its findings is erroneous, a conclusion to which the clearly erroneous standard does not even apply. It is not a duplication of the district court’s role when on review an appellate court identifies errors in the district court’s reasoning. In Anderson, the Supreme Court made it clear that to uphold a district court’s findings as not clearly erroneous, the district court’s account of the evidence in making those findings must be plausible in light of the record viewed in its entirety. See Anderson, 470 U.S. at 573-74, 105 S.Ct. at 1511-12. Anderson does not hold that a trial court’s factfinding power is limitless, nor that appellate review is nothing more than a rubber stamp of the district court’s actions.
Moreover, the lead opinion states that the Jenkins class and the KCMSD did not argue that the district court’s findings are clearly erroneous. This assertion too conveniently seeks to avoid the legal arguments made by the appellants that the district court formulated conclusions which were incorrectly based upon erroneous legal premises. See Reply and Response Brief of Appellant/Cross-Appellee Kansas City, Missouri School District at 14-24.
. The district court’s fundamental misconstruction of Milliken is illustrated by the following transcript excerpts of statements made by the district judge at trial:
As I read Milliken and its progeny, the constitutional violation must have occurred in one of the suburban districts____ Now, that’s Mil-liken, as I read it. And 80 percent of [the plaintiffs’] proof in this case so far would not comport with Milliken. Now, if I’m reading Milliken wrong, then I’m reading it wrong. Tr. 7342-43.
But I was bound by law, as I read it [when dismissing the SSDs], that you cannot require an entity to be part of the [interdistrict] remedy unless you found a constitutional violation on the part of that entity * * *. Tr. 24, 561-62.
In its June 5, 1984 opinion, the district court further stated:
Because each defendant district is independent and autonomous from the state of Missouri, and from all other school districts, none has an affirmative duty to merge with other districts. None can be held vicariously liable for the acts of the state, federal government or other entities. And, absent a finding of an interdistrict constitutional violation, none can be made a part of an interdistrict remedy.
******
Instead, plaintiffs had to prove first that the segregated schools existing before 1954 were the direct and substantial cause of blacks leaving each defendant district; second, assuming that occurred, that it had a significant segre-gative effect in the KCMSD. See, 418 U.S. at 744-45, [94 S.Ct. at 3127]; Lee, 639 F.2d at 1256 ("there must be clear proof of cause and effect and a careful delineation of the extent of the effect”). Plaintiffs' proof was weak, speculative and in any event de minimus [sic], Jenkins v. State of Mo., No. 77-0420-CV-W-4, slip op. at 12 (W.D.Mo. June 5, 1984).
Even accepting Dr. Kain’s conclusion that the present racial segregation is due to discrimination, Dr. Kain offered no evidence or opinions tending to show that any actions or inactions of any SSD contributed in any significant way to causing the racial residential separation he described. The Court finds that none of the SSDs did or failed to do anything which had any significant effect on residential patterns in the Kansas City SMSA. His testimony, therefore, does not support plaintiffs’ claims for relief against the SSDs.
Jenkins, June 5 opinion at 36.
Again, in an order issued in January, 1985, the district court stated:
It would be very beneficial in the event the Eighth Circuit Court of Appeals were to reverse this Court’s dismissal of the suburban districts. It would be useful in any effort undertaken to seek voluntary consolidation between any or all of the school districts. However, because of restrictions on this Court’s remedial powers in restructuring the operations of local and state government entities, that portion of the KCMSD plan which *698would require the consolidation of eleven suburban school districts with KCMSD goes far beyond the nature and extent of the constitutional violation this Court found existed. See, Order filed September 17, 1984. This Court found that there was no unconstitutional action on the part of any of the eleven suburban school districts nor was there evidence of constitutional violations by those school districts which had any significant segregative effect within their own districts or on the KCMSD. Milliken v. Bradley, 418 U.S. 717, 745-48 [94 S.Ct. 3112, 3127-29, 41 L.Ed.2d 1069] (1974) (where respondents, alleging the Detroit school system was racially segregated, sought the creation of a unified school district as a remedy without a finding of constitutional violations by the suburban districts and without finding significant segregative effects). "The Supreme Court in Hills v. Gau-treaux, [425 U.S. 284, 96 S.Ct. 1538, 47 L.Ed.2d 792 (1976) ] has interpreted Milliken I to mean that district courts may not restructure or coerce local governments or their subdivisions." Liddell v. State of Missouri, 731 F.2d 1294, 1308 (8th Cir.1984). Clearly a plan which would require the consolidation of suburban school districts, ones in which no constitutional violation or significant segregative effect has been found, results in a restructuring and coercion of local government which is prohibited under Milliken I.
Jenkins v. State of Mo., No. 77-0420-CV-W-4, slip op. at 2 (W.D.Mo. January 25, 1985).
. In its brief on appeal, the state notes:
Thus, in any case in which a state school system exhibits a large measure of local control, Milliken / requires that parties seeking an interdistrict remedy prove either (a) that the school district boundaries were manipulated for racial reasons or (b) that some intentionally discriminatory action by the governmental defendants — the State or the local school districts — was a "substantial cause” of "significant” current interdistrict segregation. Id.; see Hills v. Gautreaux, 425 U.S. 284, 294 & n. 11 [96 S.Ct. 1538, 1545 & n. 11, 47 L.Ed.2d 792] (1976) [Hills ] (Milliken I requires demonstration of "significant” interdistrict segregative effects caused by unlawful governmental act); Lee v. Lee County Board of Education, 639 F.2d 1243, 1254-56 (5th Cir.1981).
Brief of State Appellees/Cross-Appellants at 10.
. Judge Heaney, who authored the Morrilton opinion, now joins in this opinion’s specific conclusion that Morrilton was never intended to be limited to gerrymandered boundaries. Moreover, this court’s reliance in Morrilton on Board of School Comm’rs. clearly refutes the lead opinion's position. In Morrilton, we adopted wholly the language of the Seventh Circuit, which states:
[Sjchool officials may not maintain that their districts should be excluded from any inter-district remedy if they are found innocent of committing any constitutional violations because they should not be held responsible for the acts of the state legislators or other state subdivisions such as a local housing authority or a zoning board. The commands of the Fourteenth Amendment are directed at the state and cannot be avoided by a fragmentation of responsibility among various agents. * * * If the state has contributed to the separation of the races, it has the obligation to remedy the constitutional violations. That remedy may include school districts which are its instrumentalities and which were the product of the violation.
Board of School Comm'rs., 573 F.2d at 410 (quoted in Morrilton, 606 F.2d at 228-29) (emphasis added) (citation omitted). The lead opinion attempts to distinguish this language from Board of School Comm'rs. by asserting that the school districts in that case were "mere instru-mentalities of the state." Ante at 673. As grounds for its conclusion that the SSDs are "independent and their boundaries must be respected,” the district court stated that "to find the SSDs vicariously liable for acts of the state, the Court must make the threshhold determination that the SSDs are mere agents or arms of the state and not independent, locally autonomous entities.” Jenkins, June 5 opinion at 101-02. See also id. at 8-14. Not only does the lead opinion’s statement reflect the district court’s misconstruction of Milliken, but it overlooks the fact that an SSD may have some attributes of autonomy and yet remain an instrumentality of the state.
In its June 5 opinion, the district court found that the SSDs had more attributes of autonomy than those Michigan school districts examined in Milliken. Jenkins, June 5 opinion at 8. While it is true that the SSDs may have some authority to act independently, that authority is delegated to them by the state and the state retains ultimate authority over the SSDs' actions. The SSDs exist pursuant to provisions of the Missouri constitution, are maintained at the pleasure of the Missouri legislature, and are subject to the authority of state statutes and the rules and regulations of the Missouri State Board of Education. See, e.g., Mo. Const, arts. Ill, § 40(20); IX. Evidence introduced at trial acknowledged the authority of the Missouri General Assembly to reorganize Missouri’s schools for the purpose of achieving racial balance. See, e.g., Plaintiffs’ Exhibit 1010. The Missouri Supreme Court has found that school districts “form an integral part of the state, and constitute that arm or instrumentality thereof discharging the constitutionally [e]ntrusted governmental function of imparting knowledge and intelligence to the youth of the state,” School Dist. v. School Dist., 340 Mo. 779, 102 S.W.2d 909, 910 (1937), and that the school districts are statutory trustees for the discharge of the governmental function to apportion funds in the best interests of education which by the constitution is entrusted to the state, School Dist. No. 59 v. Maple Grove School Dist. No. 56, 359 S.W.2d 743, 747-48 (Mo.1962) (citing School Dist. v. School Dist.). Moreover, the district court explicitly concluded in its September opinion that “none of the provisions of the Constitution nor statutes of the State of Missouri would have prevented the State of Missouri from fulfilling its affirmative duty of disestablishing a dual school system subsequent to 1954.” Jenkins v. State of Mo., 593 F.Supp. 1485, 1504 (W.D.Mo.1984). The court further concluded that ”[i]t is clear that school districts in the State exist pursuant to the State Constitution,” id., and that “[t]he State executive and its agencies as well as the State’s General Assembly had and continue to have the constitutional obligation to affirmatively dismantle any system of de jure discrimination, root and branch,” id. at 1505.
. This court’s line of cases interpreting and applying the Milliken standard to encompass a wide range of constitutional violations is consistent with Justice Stewart’s restatement of the Supreme Court’s holding in his concurring opinion in Milliken, in which he stated:
Since the mere fact of different racial composition in contiguous districts does not itself imply or constitute a violation of the Equal Protection Clause in the absence of a showing that such disparity was imposed, fostered, or encouraged by the State or its political subdivisions, it follows that no inter-district violation was shown in this case.
Milliken, 418 U.S. at 756, 94 S.Ct. at 3133 (Stewart, J., concurring) (emphasis added). Justice Stewart further emphasized that [t]he Constitution simply does not allow federal courts to attempt to change that situation unless and until it is shown that the State, or its political subdivisions, have contributed to cause the situation to exist. No record has been made in this case showing that the racial composition of the Detroit school population or that residential patterns within Detroit and in the surrounding areas were in any significant measure caused by governmental activity, and it follows that the situation over which my dissenting Brothers express concern cannot serve as the predicate for the remedy adopted by the District Court and approved by the Court of Appeals.
Id. at 756 n. 2, 94 S.Ct. at 3133 n. 2 (emphasis added). Nowhere does Justice Stewart limit the definition of unconstitutional governmental activity solely to instances of gerrymandering.
. Under Missouri law, the SSDs are legally responsible for the constitutional violations of their predecessor districts. See Lewis County C-I School Dist. v. Normile, 431 S.W.2d 118, 121 (Mo.1968) (en banc); McClure v. Princeton Reorganized School Dist., 307 S.W.2d 726, 727-28 (Mo.Ct.App.1975); Lynch v. Webb City School Dist. No. 92, 373 S.W.2d 193, 200 (Mo.Ct.App. 1963). See also Taylor v. Board of Educ., 294 F.2d 36, 38-39 (2d Cir.1961), cert. denied, 368 U.S. 940, 82 S.Ct. 382, 7 L.Ed.2d 339 (1961) (school board at time of litigation held liable for the discriminatory acts of predecessor boards. Any reference to the conduct of an SSD therefore includes liability for any acts of its predecessors.
. The district court found that four of the SSDs —Center, Raytown, Grandview, and Hickman Mills — did not operate dual school systems because they had no black residents prior to 1954. However, this finding is highly questionable, especially as to Raytown, Grandview, and Hickman Mills, in light of the school records and other evidence before the district court which showed the presence of at least some black children in their predecessor districts. In any event, that no black children may have lived in these districts prior to 1954 does not exculpate these districts. The great weight of evidence at trial indicated that the fact that there were no schools for black children to attend in these SSDs deterred blacks from moving to those districts in the first place. See, e.g., Jenkins, 593 F.Supp. at 1490; testimony of Dr. James D. Anderson, an expert in the field of the history of American education, summarized in Brief of Amicus Curiae The Kansas City Branch of The National Association for the Advancement of Colored People at 3-8.
. This attempt to separate constitutional violations related to housing and the segregative policies of the various school districts ignores the district court’s undisputed finding of the "inextricable connection between schools and housing.” Jenkins, 593 F.Supp. at 1491. The constitutional violations here are so intertwined that it is impossible to separate the effects of one from another.
. But see this dissent's footnote 6, supra, criticizing the lead opinion’s conclusion that the SSDs are not instrumentalities of the State of Missouri.
. Because Missouri mandated separate schools for blacks and whites until 1954, the current disparity between the percentage of blacks attending school in the SSDs and KCMSD should raise a rebuttable presumption that the de facto segregation that currently exists is the result of the prior constitutional violations. See Swann v. Charlotte-Mecklenburg Bd. of Educ., 402 U.S. 1, 26, 91 S.Ct. 1267, 1281, 28 L.Ed.2d 554 (1971). The lead opinion argues that the Swann presumption that proof of prior de jure segregation coupled with a continuing racial imbalance establishes liability unless rebutted is inapplicable in interdistrict cases. To support this conclusion, the lead opinion notes that the Supreme Court did not apply the Swann presumption in Milliken. Ante at 670. Unlike this case, however, there was no allegation nor proof in Mil-liken that prior de jure segregation had interdis-trict effects. In any event, even if the Swann presumption does not apply to interdistrict cases, the plaintiffs presented sufficient evidence that the constitutional violation of the SSDs have current interdistrict effects.
. Other examples of KCMSD’s segregative policies include the transfer of its best and most experienced black teachers to white schools, leaving less capable teachers in black and changing schools and accelerating those schools’ transition to all-black schools. Too, as the district court found, the liberal transfer policy allowed "whites living in racially transitional neighborhoods in Kansas City to transfer * * * to whiter schools.” Jenkins, 593 F.Supp. at 1493.
. The lead opinion asserts that this statement is an example of "appellate factfinding." To the contrary, this statement is based on the district court’s own observations as to what actually occurred in the Kansas City metropolitan area due to institutional segregation. See Jenkins, 593 F.Supp. at 1493-94.
. It is conceivable that the district court found nothing inconsistent in so holding, because it labored throughout these proceedings under the miscomprehension that Milliken requires each individual SSD to be a constitutional violator. The lead opinion now urges that the August 25, 1986 opinion was merely a general order which does not contradict the district court’s earlier, more lengthy orders. Ante at 671 n. 17. This statement is at odds not only with the August order's plain language, but also with the objections to that August order made to this court by the SSDs on the grounds that the order was entered after the SSDs were no longer parties to the case and that its language was prejudicial to them.
. See infra note 18.
. The court also erred in rejecting much of the plaintiffs otherwise uncontroverted expert testimony regarding the effect of defendants’ constitutional violations on the residential choices of blacks in the Kansas City metropolitan area. Much of this testimony, given by Dr. James D. Anderson, a leading scholar and historian, was uncontradicted and based on interpretation of public records. See supra note 9. This evidence was highly probative, and it was wholly arbitrary for the court to reject it and inject its own inferences. Moreover, it is especially crucial in a case of this nature to give great consideration to the testimony of historical experts where the theory to be proved occurred long ago and few witnesses are alive to testify.
. See, e.g., 1929 Report of Missouri State Superintendent of Public Schools at 122-23 ("high school opportunity for [N]egro children is very limited” other than in St. Louis and Kansas City; "84 percent of all the high school education in the state" for blacks is provided in these cities although "less than half the [N]egro population in the state" lives there; in light of poor education for blacks in areas outside the cities, “the pronounced drift of the [N]egro population away from the farm to the city is quite understandable”); 1924 Report at 197 (”[m]any [blacks in rural areas] leave to seek better educational facilities for their children”).
. The lead opinion’s implication that interdis-trict relief here would trigger white flight to Kansas, ante at 686, is irrelevant in determining whether interdistrict relief is appropriate. See, e.g., United States v. Board of School Comm’rs., 503 F.2d 68, 80 (7th Cir.1974) (Seventh Circuit reversed district court order for interdistrict relief, but found prediction of "white flight" an unacceptable reason for failing to desegregate schools).
. As we observed in Little Rock, in Milliken it was proposed that the remedy include consolidation of one city district with fifty-three suburban districts in three counties where there was no evidence in the record indicating that the fifty-four districts were closely interrelated geographically, economically, politically or culturally as are the districts in this case. Little Rock, 778 F.2d at 429.
. Although for the reasons stated herein we stand in dissent, we nonetheless agree with Judge Ross' statement as to the likely effects of refusal by the Kansas City metropolitan school districts to participate now or in the future in any cross-district programs.