with whom Mr. Justice Powell joins, dissenting.
The school desegregation remedy imposed on the Columbus school system by this Court’s affirmance of the Court of Appeals is as complete and dramatic a displacement of local authority by the federal judiciary as is possible in our federal system. Pursuant to the District Court’s order, 42,000 of the system’s 96,000 students are reassigned to new schools. There are like reassignment of teachers, staff, and administrators, reorganization of the grade structure of virtually every *490elementary school in the system, the closing of 33 schools, and the additional transportation of 37,000 students.
It is difficult to conceive of a more serious supplantation because, as this 'Court recognized in Brown v. Board of Education, 347 U. S. 483, 493 (1954) (Brown I), “education is perhaps the most important function of state and local governments” ; indeed, it is “a vital national tradition.” Dayton Board of Education v. Brinkman, 433 U. S. 406, 410 (1977) (Dayton I); see Milliken v. Bradley, 418 U. S. 717, 741-742 (1974); Wright v. Council of City of Emporia, 407 U. S. 451, 469 (1972). That “local autonomy has long been thought essential both to the maintenance of community concern and support for public schools and to quality of the educational process,” Milliken, supra, at 741-742, does not, of course, place the school system beyond the authority of federal courts as guardians of federal constitutional rights. But the practical and historical importance of the tradition does require that the existence of violations of constitutional rights be carefully and clearly defined before a federal court invades the traditional ambit of local control, and that the subsequent displacement of local authority be limited to that necessary to correct the identified violations. “It is for this reason that the case for displacement of the local authorities by a federal court in a school desegregation case must be satisfactorily established by factual proof and justified by a reasoned statement of legal principles.” Dayton I, supra, at 410.
I think the District Court and Court of Appeals in this case did not heed this admonition. One can search their opinions in vain for any concrete notion of what a “systemwide violation” consists of or how a trial judge is to go about determining whether such a violation exists or has existed. What logic is evident emasculates the key determinants set down in Keyes v. School Dist. No. 1, Denver, Colo., 413 U. S. 189 (1973), for proving the existence and scope of a violation warranting federal-court intervention: discriminatory purpose and a causal relationship between acts motivated by such a *491purpose and a current condition of segregation in the school system. The lower courts’ methodology would all but eliminate the distinction between de facto and de jure segregation and render all school systems captives of a remote and ambiguous past.
Today the Court affirms the Court of Appeals for the Sixth Circuit in this case and Dayton Board of Education v. Brinkman (Dayton II), post, p. 526, in opinions so Delphic that lower courts will be hard pressed to fathom their implications for school desegregation litigation. I can only offer two suggestions. The first is that the Court, possibly chastened by the complexity and emotion that accompanies school desegregation cases, wishes to relegate the determination of a violation of the Equal Protection Clause of the Fourteenth Amendment in any plan of pupil assignment, and the formulation of a remedy for its violation, to the judgment of a single district judge. That judgment should be subject to review under the “clearly erroneous” standard by the appropriate court of appeals, in much the same way that actions for an accounting between private partners in a retail shoe business or claimants in an equitable receivership of a failing commercial enterprise are handled. “Discriminatory purpose” and “systemwide violation” are to be treated as talismanic phrases which, once invoked, warrant only the most superficial scrutiny by appellate courts.
Such an approach is, however, obviously inconsistent with the Dayton I admonition and disparages both this Court’s oft-expressed concern for the important role of local autonomy in educational matters and the significance of the constitutional rights involved. It also holds out the disturbing prospect of very different remedies being imposed on similar school systems because of the predilections of individual judges and their good-faith but incongruent efforts to make sense of this Court’s confused pronouncements today.1 Concepts such as *492“discriminatory purpose” and “systemwide violation” present highly mixed questions of law and fact. If district court discretion is not channelized by a clearly articulated methodology, the entire federal-court system will experience the disaffection which accompanies violation of Cicero’s maxim not to “lay down one rule in Athens and another rule in Rome.”
Yet, the only alternative reading of today’s opinions, i. e., a literal reading, is even more disquieting. Such a reading would require embracing a novel analytical approach to school segregation in systems without a history of statutorily mandated separation of the races — an approach that would have dramatic consequences for urban school systems in this country. Perhaps the adjective “analytical” is out of place, since the Court’s opinions furnish only the most superficial methodology, a framework which if it were to be adopted ought to be examined in a far more thorough and critical manner than is done by the Court’s “lick and a promise” opinions today. Given the similar approaches employed by the Court in this case and Dayton II, this case suffices for stating what I think are the glaring deficiencies both in the Court’s new framework and in its decision to subject the Columbus school system to the District Court’s sweeping racial balance remedy.
I
The Court suggests a radical new approach to desegregation cases in systems without a history of statutorily mandated separation of the races: if a district court concludes — employing what in honesty must be characterized as an irrebuttable presumption — that there was a “dual” school system at the time of Brown I, 347 U. S. 483 (1954), it must find post-1954 constitutional violations in a school board’s failure to take every affirmative step to integrate the system. Put differently, racial imbalance at the time the complaint is filed is sufficient to support a systemwide, racial balance, school busing *493remedy if the district court can find some evidence of discriminatory purpose prior to 1954, without any inquiry into the causal relationship between those pre-1954 violations and current segregation in the school system.
This logic permeates the findings of the District Court and Court of Appeals, and the latter put it most bluntly.
“[T]he District Judge on review of pre-1954 history found that the Columbus schools were de jure segregated in 1954 and, hence, the Board had a continuing constitutional duty to desegregate the Columbus schools. The pupil assignment figures for 1975-76 demonstrate the District Judge’s conclusion that this burden has not been carried. On this basis alone (if there were no other proofs), we believe we would be required to affirm the District Judge’s finding of present unconstitutional segregation.” 583 F. 2d 787, 800 (1978).
In Brinkman v. Gilligan, 583 F. 2d 243, 256 (CA6 1978), also affirmed today, this post-1954 “affirmative duty” is characterized as a duty “to diffuse black and white students” throughout the system.
The Court in this case apparently endorses that view. For the Court finds that “[e]ach instance of a failure or refusal to fulfill this affirmative duty continues the violation of the Fourteenth Amendment,” ante, at 459, and the mere fact that at the time of suit “most blacks were still going to black schools and most whites to white schools” establishes current effect. Ante, at 461.
In order to fully comprehend the dramatic reorientation the Court’s opinion thus implies, and its lack of any principled basis, a brief historical review is necessary. In 1954, this Court announced Brown I and struck down on equal protection grounds laws requiring or permitting school assignment of children on the basis of race. See also Bolling v. Sharpe, 347 U. S. 497 (1954). The question of remedy was reserved for a new round of briefing, and the following Term this Court *494remanded to the District Courts in the five consolidated cases “to take such proceedings and enter such orders and decrees consistent with this opinion as are necessary and proper to admit to public schools on a racially nondiscriminatory basis with all deliberate speed the parties to these cases.” Brown v. Board of Education, 349 U. S. 294, 301 (1955) {Brown II).
The majority concedes that this case does not involve racial assignment of students mandated by state law; Ohio abandoned any “statutory requirement or authorization to operate segregated schools” by 1888. Ante, at 455. Yet, it was precisely this type of segregation — segregation expressly mandated or permitted by state statute or constitution — that was addressed by Brown I, and the mandate of the Brown cases was that “[a] 11 provisions of federal, state, or local law requiring or permitting such discrimination must yield” to “the fundamental principle that racial discrimination in public education is unconstitutional.” 349 U. S., at 298. The message of Brown II was simple and resonant because the violation was simple and pervasive.
There were, however, some issues upon which the Brown II Court was vague. It did not define what it meant by “effec-tuat[ing] a transition to a racially nondiscriminatory school system,” id., at 301, and therefore the next 17 years focused on the question of the appropriate remedy where racial separation had been maintained by operation of state law.
The earliest post -Brown school cases in this Court only intimated that “a transition to a racially nondiscriminatory school system” required adoption of a policy of nondiscriminatory admission.2 It was not until the 1967 Term that this *495Court indicated that school systems with a history of statutorily or constitutionally mandated separation of the races would have to do more than simply permit black students to attend white schools and vice versa. In that Term, the Court had before it “freedom-of-choice” plans put forward as desegregation remedies. The factual context of the lead case, Green v. County School Board, 391 U. S. 430 (1968), is a far cry from the complicated urban metropolitan system we confront today. The New Kent County school system consisted of two schools — one black and one white — with a total enrollment of 1,300 pupils. At the time of suit a black student had *496never attended the white school or a white student the black school.
This Court found that the “freedom-of-choice” plan approved by the District Court for the desegregation of the New Kent County schools was inadequate. Noting that the “pattern of separate 'white’ and 'Negro’ schools in the New Kent County school system established under compulsion of state laws is precisely the pattern of segregation to which Brown I and Brown II were particularly addressed,” the Court observed that Brown II charged “[s]chool boards such as the respondent then operating state-compelled dual systems . . . with the affirmative duty to take whatever steps might be necessary to convert to a unitary system in which racial discrimination would be eliminated root and branch.” 391 U. S., at 435, 437-438. In the three years following court approval of the freedom-of-choice plan in New Kent County, not a single white child had chosen to attend the historically black school, which continued to serve 85% of the county’s black schoolchildren. The Oreen Court concluded that a freedom-of-choice plan, in a school system such as this and in the absence of other efforts at desegregation, was not sufficient to provide the remedy mandated by Brown II. The Court suggested zoning, i. e., some variation of a neighborhood school policy, as a possible alternative remedy.3
*497That brings the history of school desegregation litigation in this Court to The Chief Justice’s opinion in Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. 1 (1971), upon which the majority and respondents heavily rely.4 Swann also addressed school systems with a history of statutorily or constitutionally mandated separation of the races; “[t]hat was what Brown v. Board of Education was all about.” Id., at 6. Swann was an attempt to define “in more precise terms” the appropriate scope of the remedy in cases of that nature. Ibid. It simply did not attempt to articulate the manner by which courts were to determine the existence of a violation in school systems without a history of segregation imposed by statute or the state constitution.5 Certainly school systems with such a history were charged by Brown II to “effectuate a transition to a racially non discriminatory school system.” But Swann did not speak of the failure to conform to this duty as a “continuing violation.” The specific references to an affirmative duty in Swann were to the *498duty of a school board found to have overseen a school system with state-imposed segregation to put forward a plan to remedy that situation. It was in this context that the Court observed that upon “default by the school authorities of their obligation to proffer acceptable remedies, a district court has broad power to fashion a remedy that will assure a unitary school system.” 402 U. S., at 16.6
This understanding of the “affirmative duty” was acknowledged in the first case confronting a school system without a history of state-mandated racial assignment, Keyes v. School Dist. No. 1, Denver, Colo., 413 U. S. 189 (1973). There the Court observed:
“[W]e have held that where plaintiffs prove that a current condition of segregated schooling exists within a school district where a dual system was compelled or authorized by statute at the time of our decision in Brown v. Board of Education, 347 U. S. 483 (1954) (Brown I), the State automatically assumes an affirmative duty ‘to effectuate a transition to a racially nondiscriminatory school system/ Brown v. Board of Education, 349 U. S. *499294, 301 (1955) (Brown II), see also Green v. County School Board, 391 U. S. 430, 437-438 (1968), that is, to eliminate from the public schools within their school system 'all vestiges of state-imposed segregation.’ Swann v. Charlotte-Mecklenburg Board of Education, 402 U. S. 1, 15 (1971).
“This is not a case, however, where a statutory dual system has ever existed.” Id., at 200-201 (footnote omitted).
It was at this juncture that the Court articulated the proposition that has become associated with Keyes.
“Nevertheless, where plaintiffs prove that the school authorities have carried out a systematic program of segregation affecting a substantial portion of the students, schools, teachers, and facilities within the school system, it is only common sense to conclude that there exists a predicate for a finding of the existence of a dual school system.” Id., at 201.
The notion of an “affirmative duty” as acknowledged in Keyes is a remedial concept defining the obligation on the school board to come forward with an effective desegregation plan after a finding of a dual system. This could not be clearer in Keyes itself.
“[P]roof of state-imposed segregation in a substantial portion of the district will suffice to support a finding by the trial court of the existence of a dual system. Of course, where that finding is made, as in cases involving statutory dual systems, the school authorities have an affirmative duty 'to effectuate a transition to a racially nondiscriminatory school system.’ Brown II, supra, at 301.” Id., at 203.7
*500Indeed, Keyes did not discuss the complexion of the Denver school system in 1954 or in any other way intimate the analysis adopted by the Court today.8 Rather, it emphasized that the relevance of past actions was determined by their causal relationship to current racially imbalanced conditions.
Even so brief a history of our school desegregation jurisprudence sheds light on more than one point. As a matter of history, case law, or logic, there is nothing to support the novel proposition that the primary inquiry in school desegregation cases involving systems without a history of statutorily mandated racial assignment is what happened in those systems before 1954. As a matter of history, 1954 makes no more sense as a benchmark — indeed it makes less sense — than 1968, 1971, or 1973. Perhaps the latter year has the most to commend it, if one insists on a benchmark, because in Keyes this Court first confronted the problem of school segregation in the context of systems without a history of statutorily mandated separation of the races.
As a matter of logic, the majority’s decision to turn the year 1954 into a constitutional Rubicon also fails. The analytical underpinnings of the concept of discriminatory purpose have received their still incomplete articulation in the 1970’s. It is sophistry to suggest that a school board in Columbus in 1954 could have read Brown I and gleaned from it a constitutional duty “to diffuse black students throughout the . . . system” or take whatever other action the Court today thinks it should have taken. And not only was the school board to anticipate the state of the law 20 years hence, but also to have a full *501appreciation for discrete acts or omissions of school boards 20 to 50 years earlier.9
Of course, there are always instances where constitutional standards evolve and parties are charged with conforming to the new standards. But I am unaware of a case where the failure to anticipate a change in the law and take remedial steps is labeled an independent constitutional violation. The difference is not simply one of characterization: the Court’s decision today enunciates, without analysis or explanation, a new methodology that dramatically departs from Keyes by relieving school desegregation plaintiffs from any showing of a causal nexus between intentional segregative actions and the conditions they seek to remedy.
Causality plays a central role in Keyes as it does in all equal protection analysis. The Keyes Court held that before the burden of production shifts to the school board, the plaintiffs must prove “that the school authorities have carried out a systematic program of segregation affecting a substantial portion of the students, schools, teachers, and facilities within the school system.” 413 U. S., at 201 (emphasis added). The Court recognized that a trial court might find “that a lesser degree of segregated schooling . . . would not have resulted even if the Board had not acted as it did,” and “that at some point in time the relationship between past segregative acts and present segregation may become so attenuated as to be incapable of supporting a finding of de jure segregation warranting judicial intervention.” Id., at 211. The relevance of past acts of the school board was to depend on whether “segregation resulting from those actions continues to exist.” Id., at 210.10 That inquiry is not central under the approach *502approved by the Court today. Henceforth, the question is apparently whether pre-1954 acts contributed in some unspecified manner to segregated conditions that existed in 1954. If the answer is “Yes,” then the only question is whether the school board has exploited all integrative opportunities that presented themselves in the subsequent 25 years. If not, a systemwide remedy is in order, despite the plaintiff’s failure to demonstrate a link between those past acts and current racial imbalance.
The Court’s use of the term “affirmative duty” implies that integration be the pre-eminent — indeed, the controlling — educational consideration in school board decisionmaking. It takes precedence over other legitimate educational objectives subject to some vague feasibility limitation. That implication is dramatically demonstrated in this case. Both lower courts necessarily gave special significance to the Columbus School Board’s post-1954 school construction and siting policies as supporting the systemwide remedy in this case.11 They did not find- — in fact, could not have found — that the siting and construction of schools were racially motivated. As the District Court observed:
“In 1950, pursuant to a request of the then Columbus school superintendent, the Bureau of Educational Research at The Ohio State University began a comprehensive, scientific and objective analysis of the school plant needs of the school system. The Bureau studied and re*503ported on community growth characteristics, educational programs, enrollment projections, the system’s plan of organization, the existing plant, and the financial ability of the community to pay for new school facilities. Thereafter, a number of general and specific recommendations were made to the Columbus Board by the Bureau. The recommendations included the size and location of new school sites as well as additions to existing sites. The recommendations were conceived to accommodate the so-called 'community or neighborhood school concept.’ The 1950 concept was related to a distance criteria grounded on walking distance to schools as follows: % mile for elementary, iy2 miles for junior high and 2 miles for senior high students.
“The Board of Education adopted and relied upon the Bureau’s recommendations in proposing and encouraging the passage of bond issues in 1951, 1953, 1956, 1959 and 1964. School construction of new facilities and additions to existing structures were accomplished in substantial conformity with the Bureau’s periodic studies and recommendations.” 429 F. Supp. 229, 237-238 (SD Ohio 1977).
Thus, the Columbus Board of Education employed the most objective criteria possible in the placement of new schools.
Nevertheless, the District Court and Court of Appeals found that conformity with these recommendations was a violation of the Equal Protection Clause because “in some instances the need for school facilities could have been met in a manner having an integrative rather than a segregative effect.” Id., at 243.12 By endorsing this logic, the Court, as a result of its *504finding of an affirmative duty, employs remedy standards to determine the existence of post-1954 violations in school construction and ignores the previously pivotal role of discriminatory purpose.13
*505This unprecedented “affirmative duty” superstructure sits atop a weak foundation — the existence of a “dual” school system in 1954. This finding was predicated on the presence *506of four predominantly black elementary schools and one predominantly black junior high school on the “near east side of Columbus,” a then and now black residential area. The Columbus School Board at that time employed, as it does now, a neighborhood school policy. The specific Board actions that the District Court cited were racial assignment of teachers and gerrymandering along part of the border between two school districts.14 The Court concludes that these violations involved a substantial part of the Columbus school system in 1954, and invokes Keyes for the proposition that the finding of a dual school system follows “absent sufficient contrary proof by the Board, which was not forthcoming in this case.” Ante, at 458.
There are two major difficulties with this use of Keyes. First, without any explanation, the Court for the first time applies it to define the character of a school system remote in time — here 25 or more years ago — without any examination of the justifications for the Keyes burden-shifting principles when those principles are used in this fashion. Their use is a matter of “ ‘policy and fairness/ ” 413 U. S., at 209 (quoting 9 J. Wigmore, Evidence § 2486, p. 275 (3d ed. 1940)), and I think the Keyes “presumption” scores poorly on both counts when focused on a period beyond memory and often beyond *507records.15 What records are available are equally available to both sides. In this case the District Court relied almost exclusively on instances that occurred between 1909 and 1943: undoubtedly beyond the period when many Board members had their experiences with the system as students, let alone as administrators. It is much more difficult for school board authorities to piece together the influences that shaped the racial composition of a district 20, 30, or 40 years ago. The evidence on both sides becomes increasingly anecdotal. Yet the consequences of the School Board's inability to make such a showing only become more dramatic. Here violations with respect to 5 schools, only 3 of which exist today, occurring over 30 years ago are the key premise for a systemwide racial *508balance remedy involving 172 schools — most of which did not exist in 1950.16
My second concern about the Court's use of the Keyes presumption may render my first concern academic. For as I suggest in Part III below, the Court today endorses views regarding the neighborhood school policy and racially identifiable neighborhoods that essentially make the Keyes presumption irrebuttable.
II
The departure from established doctrines of causation and discriminatory purpose does not end with the lower courts’ preoccupation with an “affirmative duty” exhumed from the conduct of past generations to be imposed on the present without regard to the forces that actually shaped the current racial imbalance in the school system. It is also evident in their examination of post-1954 violations, which the Court refers to as “the intentionally segregative use of optional attendance zones, discontiguous attendance areas, and boundary changes.” Ante, at 461-462 (footnotes omitted).
As a preliminary matter, I note that the Court of Appeals observed, I think correctly, that these post-1954 incidents “can properly be classified as isolated in the sense that they do not form any systemwide pattern.” 583 F. 2d, at 805. All the incidents cited, let alone those that can meet a properly applied segregative intent standard, could not serve as the basis for a systemwide racial balance remedy.
In Washington v. Davis, 426 U. S. 229 (1976), Arlington Heights v. Metropolitan Housing Dev. Corp., 429 U. S. 252 *509(1977), and Personnel Administrator of Massachusetts v. Feeney, 442 U. S. 256 (1979), we have emphasized that discriminatory purpose as a motivating factor in governmental action is a critical component of an equal protection violation. Like causation analysis, the discriminatory-purpose requirement sensibly seeks to limit court intervention to the rectification of conditions that offend the Constitution — stigma and other harm inflicted by racially motivated governmental action — and prevent unwarranted encroachment on the autonomy of local governments and private individuals which could well result from a less structured approach.
This Court has not precisely defined the manner in which discriminatory purpose is to be proved. Indeed, in light of the varied circumstances in which it might be at issue, simple and precise rules for proving discriminatory purpose could not be drafted. The focus of the inquiry in a case such as this, however, is not very difficult to articulate : Is a desire to separate the races among the reasons for a school board’s decision or particular course of action? The burden of proof on this issue is on the plaintiffs. Washington v. Davis, supra, at 244-245; Arlington Heights v. Metropolitan Housing Dev. Corp., supra, at 270.
The best evidence on this score would be a contemporaneous explanation of its action by the school board, or other less dramatic evidence of the board’s actual purpose, which indicated that one objective was to separate the races. See Arlington Heights, supra, at 268. Objective evidence is also probative. Indeed, were it not, this case would warrant very little discussion, for all the evidence relied on by the courts below was of an “objective” nature.
But objective evidence must be carefully analyzed for it may otherwise reduce the “discriminatory purpose” requirement to a “discriminatory impact” test by another name. Private and governmental conduct in matters of general importance to the community is notoriously ambiguous, and for *510objective evidence to carry the day it must be a reliable index of actual motivation for a governmental decision — at least sufficient to meet the plaintiffs burden of proof on purpose or intent. We have only recently emphasized:
“ 'Discriminatory purpose’ . . . implies more than intent as volition or intent as awareness of consequences. . . . It implies that the decisionmaker . . . selected or reaffirmed a particular course of action at least in part 'because of/ not merely 'in spite of/ its adverse effects upon an identifiable group.” Personnel Administrator of Massachusetts v. Feeney, supra, at 279.
The maintenance of this distinction is important: both to limit federal courts to their constitutional missions and to afford school boards the latitude to make good-faith, colorblind decisions about how best to realize legitimate educational objectives without extensive post hoc inquiries into whether integration would have been better served — even at the price of other educational objectives — by another decision : a different school site, a different boundary, or a different organizational structure. In a school system with racially imbalanced schools, every school board action regarding construction, pupil assignment, transportation, annexation, and temporary facilities will promote integration, aggravate segregation, or maintain segregation. Foreseeability follows from the obviousness of that proposition. Such a tight noose on school board decisionmaking will invariably move government of a school system from the townhall to the courthouse.
The District Court in this case held that it was bound by the standard for segregative intent articulated by the Court of Appeals for the Sixth Circuit in Oliver v. Michigan State Board of Education, 508 F. 2d 178, 182 (1974):
''A presumption of segregative purpose arises when plaintiffs establish that the natural, probable, and foreseeable result of public officials’ action or inaction was an increase or perpetuation of public school segregation. *511The presumption becomes proof unless defendants affirmatively establish that their action or inaction was a consistent and resolute application of racially neutral policies.” 429 F. Supp., at 254 n. 3.
This is precisely the type of “impact” trigger for shifting the burden of proof on the intent component of an equal protection violation that we rejected in Washington v. Davis, supra. There the Court of Appeals had applied the standards of Title YII to determine whether a qualifying test for police candidates discriminated against blacks in violation of the Equal Protection Clause. According to the Court of Appeals, the plaintiffs were initially required to show disproportionate impact on blacks.17 That impact was a constitutional violation absent proof by the defendants that the test was “an adequate measure of job performance in addition to being an indicator of probable success in the training program.” 426 U. S., at 237. Put differently, the defendants were to show that the test was the product of a racially neutral policy. This Court reversed, rejecting “the view that proof of discriminatory racial purpose is unnecessary in making out an equal protection violation.” Id., at 245.
Indeed, reflection indicates that the District Court’s test for segregative intent in this case is logically nothing more than the affirmative duty stated a different way. Under the test, a “presumption of segregative purpose arises when plaintiffs establish that the natural, probable, and foreseeable result *512of public officials’ . . . inaction was . . . perpetuation of public school segregation. The presumption becomes proof unless defendants affirmatively establish that their ... inaction was a consistent and resolute application of racially neutral policies.” If that standard were to be applied to the average urban school system in the United States, the implications are obvious. Virtually every urban area in this country has racially and ethnically identifiable neighborhoods, doubtless resulting from a mélange of past happenings prompted by economic considerations, private discrimination, discriminatory school assignments, or a desire to reside near people of one’s own race or ethnic background. See Austin Independent School Dist. v. United States, 429 U. S. 990, 994 (1976) (Powell, J., concurring). It is likewise true that the most prevalent pupil assignment policy in urban areas is the neighborhood school policy. It follows inexorably that urban areas have a large number of racially identifiable schools.
Certainly “public officials’ . .. inaction ... perpetuales] ... public school segregation” in this context. School authorities could move to pairing, magnet schools, or any other device to integrate the races. The failure to do so is a violation under Oliver unless the “inaction was a consistent and resolute application of racially neutral policies.” The policy that most school boards will rely on at trial, and the policy which the Columbus School Board in fact did rely on, is the neighborhood school policy. According to the District Court in this case, however, not only is that policy not a defense, but in combination with racially segregated housing patterns, it is itself a factor from which one can infer segregative intent and a factor in this case from which the District Court did infer segregative intent, stating that “[t]hose who rely on it as a defense to unlawful school segregation fail to recognize the high priority of the constitutional right involved.” 429 P. Supp., at 258.
*513But the Constitution does not command that school boards not under an affirmative duty to desegregate follow a policy of “integration fiber alies.” If the Court today endorses that view, and unfortunately one cannot be sure, it has wrought one of the most dramatic results in the history of public education and the Constitution. A duty not to discriminate in the school board’s own actions is converted into a duty to ameliorate or compensate for the discriminatory conduct of other entities and persons.
I reserve judgment only because the Court at points in its opinion seems of the view that the District Court applied a test other than the Oliver test for segregative intent, despite the District Court’s clear indication to the contrary. 429 F. Supp., at 253-254, n. 3. In fact, in Dayton II, post, at 536 n. 9, the Court expressly rejects the Oliver test, and in its opinion in this case, ante, at 464-465, indicates that the District Court treated foreseeable effects as only another bit of evidence and finds that not incompatible with this Court’s prior cases.
“Those cases do not forbid 'the foreseeable effects standard from being utilized as one of the several kinds of proofs from which an inference of segregative intent may be properly drawn.’ [429 F. Supp.], at 255. Adherence to a particular policy or practice, with full knowledge of the predictable effects of such adherence upon racial imbalance in a school system is one factor among many others which may be considered by a court in determining whether an inference of segregative intent should be drawn.’ Ibid.”
I have no difficulty with the proposition that foreseeable effects are permissible considerations “as one of the several kinds of proofs” as long as they are not the only type of proof. Use of foreseeable effects in the latter fashion would be clearly inconsistent with Davis, Arlington Heights, and Feeney. But I do have great difficulty with this Court’s taking the above *514quotations from the District Court out of context and thereby-imputing a general test for discriminatory purpose to the District Court from a passage which in fact was part of a discussion of the probativeness of a very special kind of evidence on intent: a neighborhood school policy simpliciter.18 As far as gauging the purpose underlying specific actions is concerned, it is quite clear from its expression and application of the relevant test for intent, that the District Court looked for foreseeability per se.19
*515As such, the District Court’s treatment of specific post-1954 conduct reflects the same cavalier approach to causality and purpose that underlies the 1954 affirmative duty. That determination requires no more “omnipotence and omniscience,” ante, at 457 n. 6, than similar determinations in Dayton I, Davis, and Arlington Heights. The court found violations with respect to three optional attendance zones. The Near-Bexley zone, the only zone discussed by this Court, afforded students the option to attend schools in either one of two bordering districts. The District Court found that the zone gave white students of Bexley the opportunity to avoid attending the predominantly black schools to the east. I do not think that the District Court finding can be said to be clearly erroneous despite the lack of any direct evidence on discriminatory purpose, for the School Board did not suggest any educational justification for this zone and none is apparent. But as that court recognized, the zone is of little significance as far as the current state of segregation in the school system is concerned. “The July 10, 1972, minutes of the State Board of Education . . . appear to indicate that in 1972, there were 2d public elementary school students and two public high school students residing in the optional zone.” 429 F. Supp., at 245 (emphasis added). As of 1975, the zone has been dismantled, and the District Court clearly suggests that it does not have any current effect on the Columbus school system.20
Two other optional attendance zones were identified as offen*516sive. One existed for two years, between 1955 and 1957, and permitted students in a predominantly white neighborhood to attend the “white” West Broad Elementary School rather than the predominantly black Highland School. Like the Near-Bexley option, there is no apparent educational justification and, therefore, no grounds to upset the District Court’s finding of a violation. This optional zone afforded the District Court an excellent opportunity to probe the effects of a past violation, because in 1957 the optional zone was made a permanent part of the West Broad district. But the District Court made no findings as to the current effect of the past violation nor saw fit to hypothesize how many students might have been affected. It was clearly of the opinion that no such inquiry was necessary.
The final optional attendance zone demonstrates the influence of the “affirmative duty”- — whether the 1954 variety or that which follows from Oliver. This optional zone was also created in 1955 in roughly the same part of Columbus. It gave some students within Highland’s boundaries the option of attending the neighboring West Mound Street Elementary School. Again, the District Court found, this permitted transfer to a “whiter” school. But the District Court also found that there was a legitimate educational objective for creation of the zone: Highland was overcrowded and West Mound was under capacity. The District Court, however, concluded that the School Board’s actions were objectionable because “feasible alternatives” were available; that is, other optional attendance zones could have been drawn which would have had “an integrative effect on West Mound.” This again suggests a duty on the School Board to select the most integrative alternative.
The second set of post-1954 actions faulted by the District Court were two discontiguous attendance areas. These were situations where students in a defined geographical area were assigned to a school in a zone not contiguous with their neigh*517borhood. One zone was established in 1963 and involved about 70 students. The School Board unsuccessfully argued at trial that the children were sent to the predominantly white Moler Elementary School because the nearest school, the predominantly black Alum Crest Elementary, had no room for them. The District Court indicates that this violative condition existed until 1969, presumably because after that date the discontiguous area had a substantial black population and an integrative effect on the Moler Elementary School. Since the discontiguous area now has an integrative effect, one might ask what is its current segregative effect on the school system? Ironically, under the District Court’s reasoning, it would be a violation for the Columbus School Board to now disband the Moler Elementary discontiguous attendance area.
The second discontiguous zone existed from 1957 to 1963 and permitted students on three streets within the Heimandale Elementary District to attend the “whiter” Fornof Elementary School. The Columbus School Board “inherited” this dis-contiguous attendance arrangement when it annexed the Marion-Franklin District in 1957. Both schools at that time were at or over capacity and when a six-classroom addition was made to Heimandale in 1963, the discontiguous zone was terminated and the children assigned to Heimandale. According to the HEW Civil Rights Survey, Heimandale today is a racially balanced school. App. 747. The District Court made no findings as to the current effect of the Board’s 5-year retention of the Heimandale-Fornof arrangement.
The last discrete violation discussed by the District Court involved the Innis-Cassady alternative organizational proposals. These proposals involved an area of the Columbus school district that was annexed in 1971. The area had one school, the Cassady Elementary School, which was very overcrowded, and placing x another school in the district was a priority for the Columbus School Board in 1972. The District Court did not fault the site chosen for the second school in the old Mifflin District. However, it inferred segregative *518intent in the School Board’s decision to use a K-6 organization in both schools, rather than using K-3 organization in one school and 4-6 organization in the other and thereby drawing students from throughout the district. The District Court found that the latter would have been the more integrative alternative because of residential segregation in the district. At trial, the School Board attempted to justify its choice by pointing out that the pairing alternative would have required substantial transportation and a deviation from the standard K-6 organization employed throughout the Columbus school system. The court found “no evidence in this record” that pairing would have necessitated “substantial transportation” and that the Board had on prior occasions used a K-3 structure^ — apparently a reference to the K-3 primary center for crippled children.21
Thus, the Innis-Cassady discussion evinces this same affirmative duty to select the more integrative alternative and a consequent shift of the burden of proof to the School Board to prove that the segregative choice was mandated by other legitimate educational concerns. But under Washington v. Davis and Arlington Heights the burden is on the plaintiffs to show impact and purpose, and in a situation where there is “no evidence” in the record to prove or disprove a proffered justification for a school board decision, the plaintiffs have failed to establish a violation of their constitutional rights.
Secondly, the fact that a school board has once or twice or three times in the past deviated from a policy does not impugn that policy as a justification for a school board decision. There is no constitutional requirement of perfect consistency. Arlington Heights, 429 U. S., at 269. The fact that the Columbus School Board currently maintains a K-3 orga*519nization for crippled children hardly diminishes the Board’s interest in maintaining a standard organizational structure for traditional schools throughout the school district.22 Rather, in Arlington Heights we spoke of substantive departures from existing policy as casting light on discriminatory purpose, “particularly if the factors usually considered important by the decisionmaker strongly favor a decision contrary to the one reached.” Id., at 267.
Thus, it is clear that with respect to a number of the post-1964 actions that the District Court found to be independent violations, foreseeability was not one kind of evidence, but the whole ball game — whether the District Court thought that result dictated by the Oliver test or the post-1954 “affirmative duty” purportedly imposed as a result of pre-1954 conduct. Those findings that could be supported by the concept of discriminatory purpose propounded in Davis and Arlington Heights were not accompanied by any effort to link those violations with current conditions of segregation in the school system. In sum, it is somewhat misleading for the Court to refer to these actions as in some sense independent of the constitutional duty it suggests that the Columbus Board assumed in 1954. And, in any event, the small number of students involved in these instances could not independently support the sweeping racial balance remedy imposed by the District Court. Cf. Dayton I, 433 TJ. S. 406 (1977).
Ill
The casualness with which the District Court and Court of Appeals assumed that past actions of the Board had a *520continuing effect on the school system, and the facility and doctrinal confusion with which they went from these actions to announce a “systemwide violation” undermine the basic limitations on the federal courts’ authority. If those violations are not the product of a careful inquiry of the impact on the current school system, if they are reaction to taint or atmosphere rather than identifiable conditions that would not exist now “but for” the constitutional violation, there are effectively no limits on the ability of federal courts to supplant local authority. Only two Terms ago, in Dayton I, supra, at 420, we set out the basic line of inquiry that should govern school desegregation litigation:
“The duty of both the District Court and the Court of Appeals in a case such as this, where mandatory segregation by law of the races in the schools has long since ceased, is to first determine whether there was any action in the conduct of the business of the school board which was intended to, and did in fact, discriminate against minority pupils, teachers, or staff. Washington v. Davis, supra. All parties should be free to introduce such additional testimony and other evidence as the District Court may deem appropriate. If such violations are found, the District Court in the first instance, subject to review by the Court of Appeals, must determine how much incremental segregative effect these violations had on the racial distribution of the Dayton school population as presently constituted, when that distribution is compared to what it would have been in the absence of such constitutional violations. The remedy must be designed to redress that difference, and only if there has been a sys-temwide impact may there be a systemwide remedy. Keyes, 413 U. S., at 213.”
See also School Dist. of Omaha v. United States, 433 U. S. 667 (1977); Brennan v. Armstrong, 433 U. S. 672 (1977).
*521The District Court made no attempt to determine the incremental segregative effects of identified violations; given the absence of causality considerations in the court’s findings, it was simply not in a position to do so.23 To distinguish Dayton I, the majority relies on the District Court’s conclusion that its “finding of liability in this case concerns the Columbus school district as a whole.” 429 F. Supp., at 266. But incantation is not a substitute for analysis and the District Court’s findings and analysis do not support its conclusion.
But the majority’s opinion takes on its most delusive *522air when the Court suggests that the scope of the remedy is the Board’s own fault.
“[T]he Board was given ample opportunity to counter the evidence of segregative purpose and current, system-wide impact, and the findings of the courts below were against it in both respects.” Ante, at 468.
Specifically, the Court is alluding to the Board’s purported failure to show that the violation was not systemwide under Keyes or that a more limited remedy should have been applied under Swann. In fact, the logic of the District Court, apparently endorsed by the Court today, turns the Swann and Keyes showings into chimeras.
Once a showing is made that the District Court believes satisfies the Keyes requirement of purposeful discrimination in a substantial part of the school system, the School Board will almost invariably rely on its neighborhood school policy and residential segregation to show that it is not responsible for the existence of certain predominantly black and white schools in other parts of the school system. Under the District Court’s reasoning, as I have noted, not only is that evidence not probative on the Board’s lack of responsibility, it itself supports an inference of a constitutional violation. In addition, the District Court relied on a general proposition that “there is often a substantial reciprocal effect between the color of the school and the color of the neighborhood it serves” to block any inquiry into whether racially identifiable schools were the product of racially identifiable neighborhoods or whether past discriminatory acts bore a “but for” relationship to current segregative conditions.24
“It is not now possible to isolate these factors and draw *523a picture of what Columbus schools or housing would have looked like today without the other’s influence. / do not believe that such an attempt is required.
“I do not suggest that any reasonable action by the school authorities could have fully cured the evils of residential segregation. The Court could not and would not impose such a duty upon the defendants. I do believe, however, that the Columbus defendants could and should have acted to break the segregative snowball created by their interaction with housing. That is, they could and should have acted with an integrative rather than a segre-gative influence upon housing; they could and should have been cautious concerning the segregation influences that are exerted upon the schools by housing. They certainly should not have aggravated racial imbalance in the schools by their official actions.” 429 F. Supp., at 259 (emphasis added).
But, as the District Court recognized, other factors play an important role in determining segregated residential patterns.
“Housing segregation has been caused in part by federal agencies which deal with financing of housing, local housing authorities, financing institutions, developers, landlords, personal preferences of blacks and whites, real estate brokers and salespersons, restrictive covenants, *524zoning and annexation, and income of blacks as compared to whites." Ibid.
The Swann Court cautioned that “[t]he elimination of racial discrimination in public schools is a large task and one that should not be retarded by efforts to achieve broader purposes lying beyond the jurisdiction of school authorities. One vehicle can carry only a limited amount of baggage.” 402 U. S., at 22. Yet today the School Board is called to task for all the forces beyond its control that shaped residential segregation in Columbus. There is thus no room for Keyes or Swann rebuttal either with respect to the school system today or that of 30 years ago.
IV
I do not suggest that the inquiry required by Dayton I and Keyes is a simple one, and reviewing courts must defer to the findings of district court judges. But appellate courts also must ensure that these judges are asking themselves the right questions: it is clear in the instant case that critical questions regarding causality and purpose were not asked at all. The city of Columbus has changed enormously in the last 25 years and with it the racial character of many neighborhoods. Incidents related here may have been paved over by years of private choice as well as undesirable influences beyond the control of school authorities, influences such as poverty and housing discrimination, both public and private. Expert testimony should play an important role in putting together the demographic history of a city and the role of a school board in it. I do not question that there were constitutional violations on the part of the Columbus School Board in the past, but there are no deterrence or retribution components of the rationale for a school desegregation remedy. The fundamental mission of such remedies is to restore those integrated educational opportunities that would now exist but for purposefully discriminatory school board conduct. Because critically important questions were neither asked nor answered *525by the lower courts, the record before us simply cannot inform as to whether so sweeping a remedy as that imposed is justified.
At the beginning of this dissent, far too many pages ago, I suggested that the Court’s opinion may only communicate a “hands-off” attitude in school desegregation cases and that my concerns should therefore be institutional rather than doctrinal. School desegregation cases, however, will certainly be with this Court as long as any of its current Members, and I doubt the Court can for long, like Pilate, wash its hands of disparate results in cases throughout the country.
It is most unfortunate that the Court chooses not to speak clearly today. Dayton I and Keyes are not overruled, yet their essential messages are ignored. The Court does not intimate that it has fathomed the full implications of the analysis it has sanctioned — an approach that would certainly make school desegregation litigation a “loaded game board,” Swann, 402 U. S., at 28, but one at which a school board could never win. A school system’s only hope of avoiding a judicial receivership would be a voluntary dismantling of its neighborhood school program. If that is the Court’s intent today, it has indeed accepted the role of Judge Learned Hand’s feared “Platonic Guardians,” 25 and intellectual integrity — if not the Constitution or the interests of our beleaguered urban school systems and their students of all races — would be better served by discarding the pretextual distinction between de facto and de jure segregation. Whether the Court’s result be reached by the approach of Pilate or Plato, I cannot subscribe to it.
See Dayton Board of Education v. Brinkman (Dayton II), post, p. 542 (Rehnqtjist, J., dissenting).
Cooper v. Aaron, 358 U. S. 1 (1958); Goss v. Board of Education, 373 U. S. 683 (1963); Griffin v. School Board, 377 U. S. 218 (1964).
In discussing the Brown II mandate, this Court in Cooper v. Aaron, supra, at 7, observed:
“Of course, in many locations, obedience to the duty of desegregation would require the immediate general admission of Negro children, otherwise qualified as students for their appropriate classes, at particular *495schools. On the other hand, a District Court, after analysis of the relevant factors (which, of course, excludes hostility to racial desegregation), might conclude that justification existed for not requiring the present nonsegregated admission of all qualified Negro children.”
A similar limited expectation pervades Goss v. Board of Education, supra, where this Court invalidated court-ordered desegregation plans which permitted transfers on the basis of race. Specifically, the desegregation plan called for the redrawing of school districts without reference to race, but explicitly authorized transfers by students of one race from a school where their race was a minority to a school where their race was a majority. There was no provision for majority-to-minority school transfers. This Court objected to the explicit racial character of the transfer program.
“Our task then is to decide whether these transfer provisions are . . . unconstitutional. In doing so, we note that if the transfer provisions were made available to all students regardless of their race and regardless as well of the racial composition of the school to which he requested transfer we would have an entirely different case. Pupils could then at their option (or that of their parents) choose, entirely free of any imposed racial considerations, to remain in the school of their zone or transfer to another.” 373 U. S., at 687.
Griffin v. School Board, supra, involved a situation where a school system literally closed down its schools rather than desegregate. The decree endorsed by this Court, in the face of massive resistance, was simply an order to the school board requiring it to admit students without regard to race to a white high school and to make plans for admissions to elementary schools without regard to race.
Two other cases were handed down on the same day as Green. Raney v. Board of Education, 391 U. S. 443 (1968), involved an almost identical factual situation with a similar experience under a freedom-of-choice plan. For the same reasons that such a plan was inadequate for New Kent County, it was found inadequate for the Gould School District involved in the Raney litigation. The other case handed down with Green, Monroe v. Board of Comm’rs, 391 U. S. 450 (1968), concerned the city of Jackson, Tenn. At issue in that case was a “free-transfer” rather than “freedom-of-choice” plan. The “free-transfer” provisions were part of a court-ordered plan that essentially instituted a neighborhood school policy for the three junior high schools in the system. Any child could transfer to another school if space was available, i. e., if there were no neighborhood-zone residents to fill the spaces. This Court did not object to the *497neighborhood school policy as part of a remedy, even though some neighborhoods were racially identifiable, but it found that the effect of the free-transfer policy was to maintain the racial characters of the three junior high schools. One remained all black and another 99% white.
There were two school desegregation cases heard in this Court in the years between Swarm and Oreen. Alexander v. Holmes County Board of Education, 396 U. S. 19 (1969), reiterated that the era of “all deliberate speed” had ended. United States v. Montgomery County Board of Education, 395 U. S. 225 (1969), involved an order requiring the reassignment of some faculty and staff of the Montgomery County school system in line with numerical targets set by the District Court.
Nevertheless, the Court of Appeals refers to Swann as an opinion which “dealt more thoroughly than any other opinion of the Court with the method of proof of constitutional violations,” 583 F. 2d 787, 793 (CA6 1978), and relies on it throughout its opinion for standards of proof in determining the existence of a violation. Swarm was in fact an attempt to articulate the “equitable remedial discretion of the District Court” which admits more latitude than the standards for determining a violation. 402 U. S., at 25; see id., at 15-16. There is no “discretion” in the latter context.
Later in its opinion, the Swann Court refers to the District Court's finding, “approved by the Court of Appeals, that the school board had totally defaulted in its acknowledged duty to come forward with an acceptable plan of its own, notwithstanding the patient efforts of the District Judge who, on at least three occasions, urged the board to submit plans.” Id., at 24.
Four other cases came down the same day as Swann. One was dismissed for lack of jurisdiction, Moore v. Charlotte-Mecklenburg Board of Education, 402 U. S. 47 (1971); one upheld a declaration that a North Carolina antibusing law was unconstitutional, North Carolina State Board of Education v. Swann, 402 U. S. 43 (1971); and another remanded a remedy order for reconsideration in light of criteria laid down in Swann, Davis v. Board of School Comm’rs of Mobile County, 402 U. S. 33 (1971). The final case, McDaniel v. Barresi, 402 U. S. 39 (1971), invalidated a state-court order barring on federal grounds a formerly statutory dual system’s voluntary transition to a modified neighborhood school policy.
The point is reiterated later in the Keyes opinion.
“If the District Court determines that the Denver school system is a dual school system, respondent School Board has the affirmative duty to desegregate the entire system 'root and branch.’ ” 413 U. S., at 213.
In fact, this theory was pressed upon the Court in Dayton I, Brief for Respondents, O. T. 1976, No. 76-539, pp. 58-71; yet it was implicitly rejected in this Court’s detailed articulation of the proper approach to equal protection challenges involving school systems “where mandatory segregation by law of the races in the schools has long since ceased.” 433 U. S., at 420.
As the Court notes, incidents relied on by the District Court occurred anywhere from 1909 to 1943.
“The essential element of de jure segregation is ‘a current condition of segregation resulting from intentional state action.’ ” Washington v. Davis, 426 U. S. 229, 240 (1976) (quoting Keyes v. School Dist. No. 1, Denver, Colo., 413 U. S., at 205).
The reliance on school construction was critical. As the Court of Appeals found, the other post-1954 incidents relied on by the District Court were “isolated,” 583 F. 2d, at 805, and therefore could not have constituted a basis for a systemwide remedy. Dayton I, 433 U. S. 406 (1977). And the only other conduct arguably having systemwide implications, racial assignment of teachers, had been corrected, was not the subject of any remedial order, 429 F. Supp. 229, 238, 260 (SD Ohio 1977), and, in any event, could not itself support the systemwide remedy under the Sixth Circuit’s own precedents. Higgins v. Board of Education of City of Grand Rapids, 508 F. 2d 779 (CA6 1974); see Dayton II, post, at 536 n. 9.
Prefacing its discussion with the observation that “in some instances initial site selection and boundary changes present integrative opportunities,” 429 F. Supp., at 241, the District Court made specific findings only with respect to 2 of the 103 schools constructed between 1950 and 1975 in the Columbus school system — Gladstone Elementary and Sixth Avenue *504Elementary — 1 of which does not exist today. The sites for both schools followed recommendations by the Bureau of Education Research of Ohio State University. Ohio State University Bureau of Educational Research, The 1958-1959 Study of the Public School Building Needs of Columbus, Ohio 58 (1959) (Sixth Avenue); Ohio State University Bureau of Educational Research, The 1963-1964 Study of the Public School Building Needs of Columbus, Ohio 65 (1964) (Gladstone).
The Gladstone Elementary School opened in 1965. The “violation” inherent in that siting is described as follows by the District Court and this passage is quoted and fully adopted by the Court of Appeals.
“The need for greater school capacity in the general Duxberry area would have been logically accommodated by the construction of Gladstone north of its present location, nearer to Hudson Street. This would, of course, require some redrawing of boundary lines in order to accommodate the need for class space in Hamilton and Duxberry. If, however, the boundary lines had been drawn on a north-south pattern rather than an east-west pattern, as some suggested, the result would have been an integrative effect on Hamilton, Duxberry and the newly-constructed school.” 429 F. Supp., at 242, quoted in 583 F. 2d, at 803.
Thus, the placement of Gladstone is a violation — not because the placement was racially motivated, it was demonstrably not so — but because another site would have had a more integrative impact, and it is a violation despite the determination by the Bureau of Educational Research that objective and legitimate educational criteria militated in favor of the Gladstone site.
The secondary status of educational objectives other than integration is even more obvious in the discussion of the Sixth Avenue School where the District Court characterized the relevant inquiry as whether “the objectives of racial integration would have been better served” by a different site and different boundaries. 429 F. Supp., at 243. The Sixth Avenue School does not exist any more, and students within its old boundaries attend two neighboring, racially balanced schools.
This is explicitly recognized by the Court in Dayton II, post, at 538 (emphasis added):
“[T]he measure of the post -Brown I conduct of a school board under an unsatisfied duty to liquidate a dual system is the effectiveness, not the *505purpose, of the actions in decreasing or increasing the segregation caused by the dual system.”
But the cases relied on by the Court, ante, at 459, to establish this affirmative duty and its implications — Dayton I, Wright v. Council of City of Emporia, 407 U. S. 451 (1972), and United States v. Scotland Neck Board of Education, 407 U. S. 484 (1972) — bear absolutely no relation to the analysis in this case. The pages cited from Dayton I simply endorse a Court of Appeals’ observation that there is nothing wrong with a school board rescinding resolutions it was under no duty to promulgate; as I have indicated, the analysis set out in Dayton I is entirely inconsistent with the “affirmative duty” invoked by the courts below. See n. 8, supra. The citation to Wright is equally mysterious. The city of Emporia is located in Greensville County, Va. Up until 1968, it was part of Greensville County’s public school system. A desegregation lawsuit was initiated in 1965 and resulted in a court-ordered “freedom-of-choice” desegregation plan for the Greensville County schools, including those within the city of Emporia. After Green, the court modified its decree and ordered pairing of certain schools. The city of Emporia then announced its intention to withdraw its schools from the Greensville County school system. The District Court enjoined it from doing so because Emporia’s schools had been part of the adjudicated dual system, and the court’s decree would be frustrated by withdrawal of the Emporia schools. In contrast the instant case has nothing to do with frustrating outstanding court orders.
United States v. Scotland Neck Board of Education, supra, was a case where the United States Department of Justice had been negotiating with the County School Board of Halifax County, N. C., in an attempt to bring it into compliance with federal law. In 1965, the schools of Halifax County were completely segregated on the basis of race. An agreement was reached that was designed to make the Halifax County school system unitary by the 1969 school year. However, in 1969, the North Carolina Legislature authorized a new independent school district in the middle of Halifax County which was to be bounded by the city limits of Scotland Neck. The United States promptly filed suit seeking desegregation of the Halifax County schools and an injunction blocking Scotland Neck’s withdrawal. The District Court ordered desegregation of the Halifax County schools and enjoined creation of the independent Scotland Neck district. This Court held, quoting Wright, that if the *506Scotland Neck “ ‘proposal would impede the dismantling of a dual system, then a district court, in the exercise of its remedial discretion, may enjoin it from being carried out.’” 407 U. S., at 489. There is certainly no support in Scotland Neck for the analysis employed today, and the Court offers no explanation.
As the Court today acknowledges, Dayton II, post, at 536 n. 9, racial assignment of teachers does not make out a Keyes showing regarding racial assignment of students. And testimony on the existence of gerrymandering went little beyond the establishment of an irregular boundary line. Testimony of W. A. Montgomery, App. 389-390. Cf. Wright v. Rockefeller, 376 U. S. 52 (1964). The District Court conceded that at the time of Brown I, there was “substantial racial mixing of both students and faculty in some schools” in the Columbus system. 429 F. Supp., at 236.
“The burdens of pleading and proof with regard to most facts have been and should be assigned to the plaintiff who generally seeks to change the present state of affairs and who therefore naturally should be expected to bear the risk of failure of proof or persuasion.” E. Cleary, McCormick on Evidence 786 (2d ed. 1972).
There is a policy judgment sometimes made, which “should not be overemphasized,” id., at 787, that the facts on a particular issue are so peculiarly within the knowledge of a certain party that the burden of proof on that issue should be allocated to him. Whatever the merits of the burden-shift to the school board where contemporaneous board decisions are at issue, see Keyes v. School Dist. No. 1, Denver, Colo., 413 U. S., at 262-263 (Rehnquist, J., dissenting), they do not commend a burden-shift regarding conduct 25 or more years ago.
The Court charges that in questioning the propriety of employing the Keyes burden-shift in this case, we “claim a better grasp of the historical and ultimate facts than the two courts below had.” Ante, at 457 n. 6. But the Keyes burden-shift is not an ultimate finding of fact at all. It is a creature of this Court, brought into play by the making of only a prima facie showing, and applied in this case in a completely novel way. To criticize its use is not to upset “factfinding,” but to criticize the absence of findings of fact which have heretofore been thought necessary in order to support the sort of remedy imposed by the District Court. Its use here is surely no less a subject for this Court’s review than it was in Keyes itself.
The Columbus school system has changed dramatically in the last 25 years. The city grew from 40 square miles in 1950 to 173 square miles in 1975, and its student enrollment more than doubled. Many of the system’s schools serve areas that were undeveloped in 1950. One hundred and three new school buildings were added during this period and 145 additions were made to existing buildings. On average, over 100 new classrooms were built each year.
To add the word “foreseeable” does not change the analysis, because the police department in Davis would be hard pressed to say that the disparate impact of the examination was unforeseeable. It is well documented that minorities do not perform as well as Anglo-Americans on standardized exams — principally because of cultural and socioeconomic differences. The Davis Court implicitly recognized that the impact in that and similar cases was foreseeable. 426 U. S., at 248, and n. 14. See Personnel Administrator of Massachusetts v. Feeney, 442 U. S. 256, 278-279 (1979).
Specifically, the District Court prefaced its discussion of the neighborhood school policy with the following question:
“If a board of education assigns students to schools near their homes pursuant to a neighborhood school policy, and does so with full knowledge of segregated housing patterns and with full understanding of the foreseeable racial effects of its actions, is such an assignment policy a factor which may be considered by a court in determining whether segregative intent exists? A majority of the United States Supreme Court has not directly answered this question regarding non-racially motivated inaction.” 429 F. Supp., at 254 (latter emphasis added).
Before today, I would have thought that the question whether nonracially motivated inaction was probative on discriminatory purpose would answer itself with an emphatic “No.” We have to date indicated that only racially motivated governmental decisionmaking is addressed by the Equal Protection Clause. It was in the course of reasoning to an affirmative answer to this question that the District Court made the first observation quoted by the Court, i. e., that the foreseeable effects of nonracially motivated inaction is probative on segregative intent. And the second quotation lifts the District Court’s conclusion on this issue out of context.
“Substantial adherence to the neighborhood school concept with full knowledge of the predictable effects of such adherence upon racial imbalance in a school system is one factor among many others which may be considered by a court in determining whether an inference of segregative intent should be drawn.” Id., at 255 (emphasis added).
Thus the interesting proposition, worthy of Lewis Carroll at his best, that a lack of discriminatory purpose will not by itself support an inference of discriminatory purpose.
In its general discussion of discriminatory intent or purpose, the District Court defines the relevant test as follows:
“The intent contemplated as necessary proof can best be described as *515it is usually described — intent embodies the expectations that are the natural and probable consequences of one’s act or failure to act. That is, the law presumes that one intends the natural and probable consequences of one’s actions or inactions.” Id., at 252.
See id., at 253-254, n. 3.
Id., at 245:
“The Court is not so concerned with the numbers of students who exercised or could have exercised this option, as it is with the light that the creation and maintenance of the option sheds upon the intent of the Columbus Board of Education.”
There were apparently only two other instances where the Columbus School Board has had K-3 primary units and both of those were to supplement overcrowding in the lower grades of K-6 home schools. Id., at 249.
There is substantial discussion in the District Court’s opinion about various groups that gave the Columbus School Board notice that certain decisions would have a segregative rather than integrative impact. Id., at 255-256. But notice in and of itself .only goes so far as to establish foreseeability, and foreseeability itself is not the ultimate fact in issue if we continue to adhere to Davis and Arlington Heights.
Dayton I was handed down after the liability phase of this ease. It was brought to the District Court’s attention while it was considering the remedy, and the District Court dismissed it as simply reiterating the maxim that “the nature of the violation determines the scope of the remedy.” Certainly Dayton I was a much more precise articulation of what implementing that maxim entailed than is found in this Court’s prior cases. And the Court of Appeals’ explanation of “incremental segregative effect” in this case communicates no clear conception of the type of inquiry into causation that Dayton I requires.
“It is clear to us that the phrases ‘incremental segregative effect’ and ‘systemwide impact’ employed in the Dayton case require that the question of systemwide impact be determined by judging segregative intent and impact as to each isolated practice, or episode. Each such practice or episode inevitably adds its own ‘increment’ to the totality of the impact of segregation. Dayton does not, however, require each of fifty segrega-tive practices or episodes to be judged solely upon its separate impact on the system. The question posed concerns the impact of the total amount of segregation found — after each separate practice or episode has added its ‘increment’ to the whole. It was not just the last wave which breached the dike and caused the flood.” 583 F. 2d, at 813-814 (emphasis in original).
In Brinkman v. Gilligan, 583 F. 2d 243, 257 (CA6 1978), the court’s description becomes metaphysical:
“The word ‘incremental’ merely describes the manner in which segregative impact occurs in a northern school case where each act, even if minor in itself, adds incrementally to the ultimate condition of segregated schools. The impact is ‘incremental’ in that it occurs gradually over the years instead of all at once as in a case where segregation was mandated by a state statute or a provision of a state-constitution.”
This empirical observation was not the product of evidence about Columbus, but general opinions expressed by two experts, Dr. Karl Taeuber and Martin Sloane; the latter testified on federal housing policy in the United States. As Mr. Justice Powell has noted, experts have found that residential segregation exists “ 'regardless of the character of *523local laws and policies, and regardless of the extent of other forms of segregation or discrimination/ ” Keyes v. School Dist. No. 1, Denver, Colo., 413 U. S., at 223 n. 9 (concurring in part and dissenting in part) (quoting Dr. Taeuber).
Dr. Taeuber credited residential segregation to economics, choice, and discrimination. In the latter category he included racially motivated site selection in public housing and urban renewal programs, restrictive covenants in housing deeds, lending policies of financial institutions, practices of the real estate industry, and zoning policies. Entering into all of this in some unspecified manner is the influence of school attendance zones. Testimony of Dr. Karl Taeuber, App. 280-311.
L. Hand, The Bill of Rights 73 (The Oliver Wendell Holmes Lectures, 1958):
“For myself it would be most irksome to be ruled by a bevy of Platonic Guardians, even if I knew how to choose them, which I assuredly do not. If they were in charge, I should miss the stimulus of living in a society where I have, at least theoretically, some part in the direction of public affairs.”