concurring in part and dissenting in part.
I.
I agree with much of the Court’s able opinion. In particular, I approve completely of its decision not to order consolidation of the three school districts now operating in Pulaski County, Arkansas. Consolidation would mean destruction of three popularly governed units of local government, and substitution in their stead of one judicially created and judicially supervised school district. Such a remedy is well within the judicial power of the United States, and I should not hesitate to support it upon proper proof, but the proof here is insufficient for several reasons, the most important of which is that the remedy of consolidation “exceeds the scope of the [parties’ constitutional] violations.” Ante at 434. *437Consolidation is a drastic step that should be reserved for clearer cases.
Having rejected consolidation, the Court proceeds to analyze the record and set out a detailed remedial decree, to be administered by the District Court on remand. The relief ordered today differs greatly from that ordered by the District Court. If we are not prepared to affirm what that court has done, we should remand this case for further findings and a detailed remedial decree. Although we have power to modify a decree at the appellate level, it is unwise to exercise that power. The District Court (though we are today disagreeing with some of its conclusions) is presided over by a scholarly and distinguished judge. That court, not this one, is in the best position to write a decree. Instead, a decree today springs full-grown from the brow of this Court, a decree that will, I dare say, startle all the parties to this case, including even those (if there are any) who like what they see.
Since the Court has decided to award detailed relief at the appellate level, however, it is appropriate for me to indicate in what respects I agree with its opinion. I agree that the District Court’s findings of intradistrict violations on the part of the North Little Rock School District (NLRSD) and the Pulaski County Special School District (PCSSD) are not clearly erroneous and should be affirmed. These violations should be corrected. Moreover, the Court properly declines to change the boundaries of NLRSD. Its constitutional defaults have not been shown to have any significant current interdistrict effect. It is also appropriate to order compensatory and remedial education programs for the four virtually all-black schools that we allowed in the Little Rock School District (LRSD) in Clark v. Board of Educ. of the Little Rock School District, 705 F.2d 265 (8th Cir. 1983). (A similar remedy might also be in order for some racially identifiable schools in NLRSD and PCSSD.) The State of Arkansas should pay for these programs. The State’s long-continued violation of the Fourteenth Amendment has played a significant part in bringing about this intradistrict condition of racial isolation.1 Some other aspects of the Court’s remedy, for example, voluntary transfers, either intradistrict or interdistrict, of students from schools where they are in a racial majority to those where they are in a minority, seem unobjectionable.
The Court directs that the boundary between LRSD and PCSSD be adjusted so that all land within the City of Little Rock shall be assigned to LRSD. It also directs the re-transfer of the Granite Mountain area to PCSSD. And, wisely, the Court’s opinion leaves it open to any party, on remand, to move the District Court to make different boundary-line adjustments, so long as they have substantially the same impact on the student populations of each district. I concur in the result reached by this portion of the Court’s opinion, though for reasons somewhat different from those it gives. In my view, PCSSD’s constitutional violations, when considered as a whole, have had some interdistrict effect, and the boundary changes ordered by this Court are a fair approximation of the measures necessary to undo that effect. There is necessarily some imprecision in this reasoning, and it rests as much on inference as-on direct evidence, but it is not unfair for the risk of erroneous decisionmaking that this kind of imprecision creates to fall, at least in part, on those who have violated the Constitution. In reaching this conclusion I am heavily influenced by expert testimony that the District Court believed, and that, accordingly, we are also obliged to accept under the clearly-erroneous rule.
*438From the remaining features of the Court’s remedy, especially its imposition of large financial responsibility on the State of Arkansas for the construction and operation of magnet schools, I respectfully dissent.
II.
One fact stands out after a reading of the District Court’s and this Court’s opinions: LRSD has more black students than either NLRSD or PCSSD. In the school year 1983-84, LRSD’s enrollment of 19,052 was about 69% black and 31% white, Tr. 1448, while PCSSD’s enrollment of 27,839 was about 22% black and 78% white, PCSSD X 64, Table 1. LRSD’s black percentage has been growing steadily, and one senses that the major impetus behind the District Court’s decision to order consolidation is a determination not to permit LRSD to become all black, or virtually so. As a policy matter, I agree that such a result is desirable. An all-black district may have problems raising adequate funds from the property tax, since most voters in the district will still be white, whatever the makeup of the public schools’ student body. It is also true that both black and white students benefit, socially and educationally, from exposure to each other.2 These facts, certainly relevant in a legislative sense, are less directly so in the present judicial context. Our task as judges is not to force these school districts to do what we think is right or socially good, but to apply the law to the facts and announce the result, whatever it may be.
Analysis must start with the governing legal standard laid down by the Supreme Court. It is stated in Milliken v. Bradley, 418 U.S. 717, 94 S.Ct. 3112, 41 L.Ed.2d 1069 (1974) (Milliken I.):
The controlling principle consistently expounded in our holdings is that the scope of the remedy is determined by the nature and extent of the constitutional violation. Swann, 402 U.S., at 16 [91 S.Ct. at 1276]. Before the boundaries of separate and autonomous school districts may be set aside by consolidating the separate units for remedial purposes or by imposing a cross-district remedy, it 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. Thus an inter-district remedy might be in order where the racially discriminatory acts of one or more school districts caused racial segregation in an adjacent district, or where district lines have been deliberately drawn on the basis of race. In such circumstances an interdistrict remedy would be appropriate to eliminate the interdistrict segregation directly caused by the constitutional violation. Conversely, without an interdistrict violation and interdistrict effect, there is no constitutional wrong calling for an interdistrict remedy.
418 U.S. at 744-45, 94 S.Ct. at 3126-28. The Supreme Court also tells us that no particular degree of racial balance is required by the Constitution, id. at 740, 94 S.Ct at 3125; that “[n]o single tradition in public education is more deeply rooted than local control over the operation of schools,” id. at 741, 94 S.Ct. at 3125; that “[t]he constitutional right of the Negro respondents residing in Detroit is to attend a unitary school system in that district,” id. at 746, 94 S.Ct at 3128; and that “[t]he suggestion ... that schools which have a majority of Negro students are not ‘desegregated’ ... finds no support in our prior cases,” id. at 747 n. 22, 94 S.Ct. at 3128 n. 22.
I also find significant the Supreme Court’s summary of the reasoning of the *439lower courts whose decisions it was reviewing:
Viewing the record as a whole, it seems clear that the District Court and the Court of Appeals shifted the primary focus from a Detroit remedy to the metropolitan area only because of their conclusion that total desegregation of Detroit would not produce the racial balance which they perceived as desirable. Both courts proceeded on an assumption that the Detroit schools could not be truly desegregated — in their view of what constituted desegregation — unless the racial composition of the student body of each school substantially reflected the racial composition of the population of the metropolitan area as a whole.
418 U.S. at 739-40, 94 S.Ct. at 3124-25.
A.
Applying this standard, I look first at the constitutional violations attributed to PCSSD. That such violations have occurred, both before and after the desegregation decree entered against PCSSD in the Zinnamon case, I do not doubt. Perhaps most shocking is the fact that current PCSSD board members, far from being familiar with the Zinnamon decree, had not even read it when they testified in the District Court. But the question for present purposes must be, what is the current interdistrict effect of these violations? See Goldsboro City Bd. of Educ. v. Wayne County Bd. of Educ., 745 F.2d 324, 330-31 (4th Cir.1984); Lee v. Lee County Board of Education, 639 F.2d 1243, 1260 (5th Cir. 1981). Have they caused more white children to come to PCSSD schools, or black children to leave or avoid them, than would otherwise have been the case? If so, to what extent? These questions must be answered, else the remedy will not fit the violation, nor the punishment fit the crime.
1. A great deal of stress is laid upon the fact that, before the Brown decision and for a time thereafter, the education provided by PCSSD for black children was grossly inferior to that provided for white children in PCSSD and to that provided for all children, black and white, in LRSD. As a consequence, some black children came to LRSD to go to school who would not otherwise have been there. But what current effect is this movement, much of which dates from 50 years ago, having? If black students came to LRSD for an education, and then went back home or elsewhere to work and raise their families, obviously their migration would not now be producing any current effect on the racial character of the LRSD student body. It is true that if black parents moved to LRSD and remained there, their descendants might now be attending LRSD schools. The Court appears to have this sort of movement in mind when it says that “[s]ome black families moved from the county to Little Rock because of the disparities in educational opportunities,” ante at 418, but the record reference cited for this statement, J.D.R. 915-19, in fact contains no support for it.3 The transfers that did occur appear to be principally of students moving into LRSD, without their parents, to live with a relative. One of LRSD’s own witnesses testified, Tr. 116, that this kind of movement of students into LRSD ceased in the 1950’s, when state laws requiring school children to be domiciled in the district where they were going to school began to be strictly enforced.4 In addition, *440any movement of black families into Little Rock that did occur must have been, to some extent, simply a part of the larger phenomenon of poor people leaving the farm-to seek opportunity in the city.
2. The Court stresses that historically, that is, when PCSSD was first formed, it was the intention of LRSD and PCSSD to expand the boundaries of LRSD pro tanto every time the City of Little Rock annexed additional territory, so that the City and LRSD would continue to be coterminous. This intention has not been adhered to: the City has annexed a good deal of territory that has remained within PCSSD and not been transferred to LRSD by “deannexation.” But only if PCSSD has declined to transfer territory to LRSD for racial reasons, in order to keep itself “white” and LRSD “black,” would this failure to “deannex” justify interdistrict relief. (No one claims that school-district lines were drawn initially for racial reasons: back in 1927, when PCSSD was formed, segregation was not thought to be unconstitutional, and no one needed to gerrymander school-district boundaries to preserve it. The claim is, rather, that the PCSSD-LRSD line was maintained for racial reasons.)
There have been eight separate transfers of territory from PCSSD to LRSD (and apparently none from LRSD to PCSSD). Of these eight transferred areas, seven have been predominantly white. Tr. 948-49. This is hardly the action of a school district seeking to maintain its “whiteness.” (The exception is the Granite Mountain area, deannexed in 1953, of which I shall speak hereafter.) The Court’s point, though, is a bit different: it charges that when it became clear that some real desegregation was going to take place, the boundaries hardened. Dr. Robert A. Dentler, LRSD’s principal expert witness, made the same point. “[T]he County I found had decided in its Board of Directors to make a formal policy of no further deannexations in 1968 ....” Tr. 343. This new policy, the Court now infers, was based on racial animus, a desire to keep the black percentage in PCSSD down.
In fact, the PCSSD Board voted, on May 14, 1968 (and LRSD now concedes this) in favor of the concept of consolidation with LRSD. Floyd Parsons, Superintendent of LRSD from 1962 to 1971, confirms that during his time in office PCSSD, the poorer district, consistently sought consolidation. It was LRSD that opposed it, and not for racial reasons, either, but simply because it did not want to take on the additional financial responsibility of educating PCSSD’s students. Tr. 1131-32 (testimony of Mr. Parsons, called by LRSD).5 It is true, therefore, that no deannexations have taken place since 1968, but to blame this on PCSSD’s desire not to increase its black student percentage is not plausible.
3. In many other respects, however, PCSSD has fallen short of its constitutional obligations, or at least the District Court has not clearly erred in so finding. It cooperated with LRSD and the State in transferring to LRSD the racially segregated Granite Mountain housing project (to be discussed in more detail later). It is imposing upon black students an unfair proportion of the burden of busing for purposes of desegregation, it is not meeting its goals for the hiring of black teachers, it is assigning black students disproportionately to the classification of educably mentally retarded, and it has failed to comply with requirements in the Zinnamon decree that a Bi-Racial Committee be established and that two black citizens serve as ex officio members of the school board. And, perhaps most important for present purposes, PCSSD has, in violation of the Zinnamon decree, located new schools in white neighborhoods or in places inconvenient to black students, and maintained a number of schools whose racial makeup falls outside the limits specified by the decree. These factors, especially the school-siting decisions, naturally affect the movement of *441students and families. PCSSD has thus violated not only the Zinnamon decree but also the Supreme Court’s direction in the Swann case, 402 U.S. at 20-21, 91 S.Ct. at 1278-79, that new schools not be located “in the areas of white suburban expansion, farthest from Negro population centers.”
I believe these factors are having a substantial current interdistrict effect. The record contains expert testimony, and it is not implausible, that these constitutional violations, considered together, are making PCSSD “whiter” and LRSD “blacker” than they otherwise would have been. I have indicated why I do not believe that PCSSD’s violations have actually caused any substantial degree of white movement from LRSD to PCSSD, or of black movement out of PCSSD. But another kind of movement — that of families coming from outside the entire Pulaski County area— has, I believe, been substantially influenced. (This sort of movement was referred to by one of the experts as “white overflight.”) The pro-white emanations that PCSSD has given off over a period of years, if I may use such a metaphor, have, it seems, been a substantial factor attracting white parents, especially since those parents could, as the lines are now drawn, move into PCSSD without being outside the City of Little Rock. The boundary change ordered by the Court (making LRSD and the City coterminous) would make LRSD 60% black and 40% white, ante at 419, instead of 70% black and 30% white. This 10% change seems a fair approximation of what the racial percentages would have been absent the influence of PCSSD’s violations. I therefore concur in the Court’s decision to adjust the LRSD-PCSSD boundary line to this extent.
B.
The Court holds that the State of Arkansas has committed constitutional violations that are producing substantial interdistrict effects. It therefore awards interdistrict relief against the State, in the person of the State Board of Education. I quite agree that the State of Arkansas has been, in this field, a persistent violator of constitutional rights. I cannot agree that these violations (with one exception) are responsible for the racial disparity now existing between PCSSD and LRSD, or that they justify (again with an exception) interdistrict relief against the State.
1. The Court recounts in detail the manifold sins and omissions of the State of Arkansas in this field. There is no point in denying the history set out in the Court’s opinion. In particular, for many years the State, although professing adherence to the “separate but equal” doctrine that was then the law of the land, in fact maintained schools that were separate and unequal, and the black people of the State bore the brunt of this inequality. Furthermore, from 1954 on, the Executive and Legislative Branches of State government set their faces like a flint against the law, covering themselves and the State with dishonor.6 But what is the present legal relevance of these facts? To the extent that any individual school district (including the three in Pulaski County) is not fully desegregated, the State is at least partly responsible and should pay the price. It has been 31 years since Brown was decided, but centuries of inequality are not so soon dissipated. There is no showing whatever, however, and no claim, that the State has caused any school-district boundary lines to be drawn or maintained for racial reasons. Much of the history that the Court details, therefore, is simply irrelevant to the question of interdistrict relief in the present case.
2. The point is made that the State Board of Education is, by statute, given extensive powers, including general supervision over all public schools in the state, Ark.Stat.Ann. § 80-113, and the approval of plans and expenditures of public-school *442funds for new school buildings, Ark.Stat. Ann. §§ 80-113, 80-3506. These statutes have never been interpreted to give the State Board of Education the sort of wide-ranging supervisory power this Court attributes to it. Instead, local school boards, except for certain legal and financial aspects of their operations, have been almost completely autonomous. See Ark.Stat. Ann. § 80-509, listing in comprehensive detail the powers of local boards. Decisions, for example, as to where to locate schools have always been treated as the prerogative of local boards. They have chosen the sites, Tr. 776, and the State Board of Education has never claimed the authority to overrule a district’s decision on where to build a school, Tr. 788. Rather, the statement in § 80-113 that the State Board shall “approve plans and expenditures of public school funds for all new school buildings” has been interpreted to authorize the State Board to review local decisions only to make sure that plans meet recognized construction standards and that proposed methods of financing are legally and fiscally sound under the statutes of the State limiting school districts’ bonded debt. Tr. 775. Both the State Board and local school districts have so construed the statute at least since 1931. There is no evidence that the State Board has ever purported to review school-siting decisions, either for desegregation or for any other purposes. It is therefore unfair to blame the State for PCSSD’s violations of the siting provisions of the Zinnamon decree.
It is true, as the Court says, that the State Board of Education’s efforts to assist and encourage desegregation have been too little and too late (though I suspect that it is entitled to somewhat more credit than the Court gives it).7 But again, what is the relevance of this fact to the specific inter-district relief contended for in the present case? If the State Board of Education had diligently fulfilled its duty to encourage desegregation, would the boundary line between LRSD and PCSSD be located in a different place from where it now is? Would the racial distribution of students between those two districts be different from what it now is? I do not believe that the record supports any definite answer to these questions. I repeat that the State’s defaults would fully justify compelling it to participate in an miradistrict remedy. But that is not what this case, at least primarily, is about.
3. There are certain specific respects, however, in which the State of Arkansas, with a racially discriminatory motive, actually assisted in the movement of school children across district lines. During the school year 1958-59, the schools in LRSD were closed, and many children from Little Rock attended segregated schools in PCSSD. The State paid at least part of the cost of these transfers, and I am willing to assume that many more white students than black benefited from this action. Shameful as it was, I cannot see that this episode has any continuing, current effect on the distribution of students as between LRSD and PCSSD. The LRSD schools reopened in the fall of 1959, and there is no evidence that students who attended school elsewhere in 1958-59 did not return to LRSD when they could. I would, however, on the basis of this history, agree that the State should pay for any voluntary majority-to-minority transfers between PCSSD and LRSD. That would be a fair recompense for what it did in the late fifties.
The Court suggests that the racial turmoil created by the State in LRSD in 1957 and the years immediately following has increased the percentage of black students in the district. It says, for example, that *443“the active intervention of the state was a central factor in delaying desegregation of the Little Rock schools until 1973, and in contributing to the increasing concentration of blacks in LRSD.” Ante at 417. With the first part of this statement I can agree completely, but the second part seems to me a non sequitur. The idea that “state-created racial turmoil in LRSD in the 1950’s fostered substantial white flight from LRSD to PCSSD and NLRSD,” ante at 417 n. 8, seems completely counterintuitive. It would be much more plausible to infer that the State’s efforts to maintain segregation in LRSD made it more likely for whites, once the schools had been reopened, to remain there. Even were it correct that pro-segregation turmoil of the late 1950’s somehow fostered white flight, this phenomenon ended long before the 1973 implementation of desegregation and could at most account for the increase in the percentage of black students to 48%, which was the black percentage in LRSD in 1973. See Clark v. Board of Educ. of the Little Rock School District, 705 F.2d 265, 267 (8th Cir.1983).
4. In one other respect, however, I believe the Court properly attributes interdistrict liability to the State. Housing authorities are creatures of the State, existing by virtue of statute, and the Little Rock Housing Authority clearly maintained and fostered racial segregation. In some cases, it might be unfair to award interdistrict relief against school districts on the basis of housing violations. A “school case, like a vehicle, can carry only a limited amount of baggage. Swann, 402 U.S. at 24 [91 S.Ct. at 1280].” Bradley v. School Board of the City of Richmond, Virginia, 462 F.2d 1058, 1066 (4th Cir.1972), aff'd by an equally divided Court, 412 U.S. 92, 93 S.Ct. 1952, 36 L.Ed.2d 771 (1973). But here, PCSSD, LRSD, and the State legislature all cooperated with the Little Rock Housing Authority in respect of the Granite Mountain Housing Project, a segregated black project constructed in 1953. At that time, territory in which the housing project was to be located was transferred by a special act of the General Assembly from PCSSD to LRSD. The school districts and the State were thus directly involved in a transfer of territory on which a segregated housing project was to be built, a fact that they must have known.
This was a clear interdistrict violation, and an appropriate remedy should be devised to cure it. The Court, ante at 435, directs that the Granite Mountain area be retransferred to PCSSD, and leaves to the District Court on remand to determine “the precise boundaries of ... the area that was impacted by the 1953 deannexation of land from PCSSD to LRSD.” Ibid. I agree that this remedy, or some substantial equivalent to be selected by the District Court, see ibid., is appropriate.
III.
In sum, this Court properly affirms, as not clearly erroneous, the District Court’s findings of intradistrict violations on the part of PCSSD. These violations, as well as those committed by NLRSD, should be corrected. In the main, intradistrict relief, in which the State Board of Education should be made to share, should be adequate for this purpose. I also believe that an interdistrict violation by PCSSD, LRSD, and the State has been made out in respect of the location of the Granite Mountain Housing Project, and that PCSSD’s other violations justify the boundary change ordered by the Court. As to the State Board of Education, however, I would not grant any interdistrict relief, except with respect to the funding of voluntary student transfers and the retransfer of the Granite Mountain area. From the extensive additional relief granted against the State, and from the remaining remedial details ordered by this Court, I respectfully dissent.
. The State argues that we cannot require it to spend more money in one school district than another, because to do so would conflict with a recent opinion of the Supreme Court of Arkansas requiring, under the State Constitution, substantially equal per-pupil funding throughout the State, DuPree v. Alma School Dist. No. 30, 279 Ark. 340, 651 S.W.2d 90 (1983), and with a statute implementing this opinion, Ark.Stat.Ann. §§ 80-850.10 — 80-850.22. This argument is insubstantial. Under the Supremacy Clause, U.S. CONST. Art. VI., cl. 2, the Fourteenth Amendment overrides any inconsistent state statute or constitutional provision.
. The proposition that all-black schools or classrooms are necessarily educationally inferior, however, is quite a different thing, and I do not subscribe to it. The "blacker” LRSD, ironically, appears by all accounts to produce more scholars of note and to offer a broader selection of courses, than the "whiter" districts with which it wishes to merge.
. Some white students also transferred to LRSD seeking a "city school education." PCSSD X 51, at p. 62; Tr. 118 (LRSD’s witness); PX 36. It was not just black schools in LRSD that were superior. All schools there were regarded as better, and this is hardly sinister or surprising, since Little Rock is the only true urban center in Arkansas and is much richer than many of the other school districts.
. The Court seems to agree that significant interdistrict movement of students stopped 20 years ago. Ante at 415. And even in the days when it was occurring, it was not simply a matter of black students transferring into LRSD. Between 1953 and 1963 more whites than blacks transferred from PCSSD to LRSD. PX 36. Some of the specific figures are instructive: In 1953-54, 47 black and 223 white children transferred from PCSSD to LRSD; in 1956-57, 42 black and 254 white; and in 1959-*44060, 34 black and 363 white. Tr. 133-34 (LRSD’s witness).
. Mr. Parsons also testified that he knew of no movement of white students from LRSD to PCSSD. Tr. 1142.
. On the other hand, the whole picture, fairly considered, is not so gloomy as the Court implies. Plaintiffs own expert witness on the history of school desegregation in Pulaski County testified that "Little Rock a few years after 1959 was far more integrated ... than many cities in the North where I grew up.” Tr. 103.
. In 1966, for example, the State Board did create a specific position to work with local boards in the desegregation process. This position was paid for out of the State’s own funds, not federal funds. Tr. 784-785. The State did not apply for federal desegregation funds. Instead, the Arkansas Technical Assistance Center, a private organization sponsored by Ouachita Baptist University, applied for and received Title V federal funds to assist school districts in desegregating. “The decision was made ... that those funds could be expended probably more efficiently if it were in any agency ... that was not subject to politics and pressures.” Tr. 804. The State Board of Education "cooperated very closely with the Center at Ouachita.” Tr. 805.