FLETCHER, Circuit Judge:
Plaintiffs, parents of all Spanish-surnamed children attending school in the San Jose Unified School District (the District), appeal the district court’s ruling that the District did not act with segregative intent in maintaining ethnically imbalanced schools. After a careful review of the record, we conclude that the district court’s finding that the defendants acted without segregative intent was clearly erroneous. Accordingly, we reverse.
I. BACKGROUND.
The Board of Education of the San Jose Unified School District (the Board) has been aware since at least 1962 that its schools are segregated. The district is long and narrow: about sixteen miles long and from one and one-half to four miles wide. Hispanics are concentrated in the [662]*662northern downtown area while the southern, suburban neighborhoods are largely Anglo. In 1973, Spanish-surnamed students made up 24.6 percent of the population of the district. In the southern schools, only .07 percent of the students were Spanish-surnamed, while in the northern schools 78.8 percent were Spanish-surnamed. State law requires the Board to take affirmative steps to desegregate its schools. The Board made repeated announcements endorsing the concept of desegregation. Nonetheless, between 1962 and 1973, the school board did nothing to alleviate the imbalance. Indeed, the Board took many steps that maintained segregation and rejected available, less segregative courses of action.
In 1971, the plaintiffs filed a class action on behalf of all Spanish-surnamed students enrolled in the District and their parents. The complaint charged that defendants were operating a segregated public school system in violation of the Fourteenth Amendment and sought desegregation of the school district. After trial to the court, the district court found that the school district was racially imbalanced and that defendants had maintained the imbalance. The court, however, denied remedial relief because it decided that defendants had acted without segregative intent. Diaz v. San Jose Unified School District, 412 F.Supp. 310 (N.D.Cal.1976) (Diaz I).
On appeal, this court vacated and remanded for reconsideration. Diaz v. San Jose Unified School District, 612 F.2d 411, 416 (9th Cir.1979) (Diaz II). We found that “[a]n inference of segregative intent arose from the appellants’ proof,” Id. at 415. The district court had, however, impermissibly concluded that the school district’s racially neutral neighborhood school policy “constituted either a complete defense to a charge of segregative intent, or, completely dispelled the inferences of segregative intent that flowed from the appellants’ proof.” Id. We held that “[sjtrict adherence to a neighborhood policy is no more than circumstantial evidence that bears upon the existence or non-existence of segregative intent. The existence of such a policy is not enough to prove the absence of segregative intent.” Id. at 416.
On remand, the district court required rebriefing and argument but took no new evidence. Once again, the court found the evidence insufficient to support a finding of segregative intent, holding that “the adherence to a neighborhood school policy by the school officials in the [San Jose Unified School District] was not a device for maintenance of segregated schools, although that effect was clearly foreseeable and was clearly foreseen by the school officials.” Diaz v. San Jose Unified School District, 518 F.Supp. 622, 644 (N.D.Cal.1981) (Diaz III). Plaintiffs appeal.
II. DISCUSSION.
A. The Need to Prove Segregative Intent.
The Board admits that it has maintained ethnically imbalanced schools and has omitted courses of action that would have reduced that imbalance. Plaintiffs presented overwhelming evidence that the Board’s actions “knowingly perpetuated ethnic imbalance in the schools.” Diaz III, 518 F.Supp. at 642. Nonetheless, de facto segregation in the schools does not, without more, prove a violation of the Fourteenth Amendment. The plaintiffs must prove not only that the defendants’ actions created or maintained racial or ethnic imbalance in the schools, but also that those actions were motivated by segregative intent. See Keyes v. School District No. 1, 413 U.S. 189, 208, 93 S.Ct. 2686, 2697, 37 L.Ed.2d 548 (1973); Diaz I, 412 F.Supp. at 328-29.
Ordinarily, only circumstantial evidence is available to establish segregative intent. Evidence of the discriminatory impact of decisions is one sort of circumstantial evidence supporting an inference of segregative intent.
The impact of the official action — whether it “bears more heavily on one race than another” ... — may provide an important starting point. Sometimes a clear pattern, unexplainable on grounds [663]*663other than race, emerges from the effect of the state action even when the governing legislation appears neutral on its face.
Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252, 266, 97 S.Ct. 555, 564, 50 L.Ed.2d 450 (1977). Proof that the segregative effects of a decision were foreseeable is not to be taken, as a general proposition, to make out a prima facie case of segregative intent, nor does it routinely shift the burden of persuasion to the defendants. See Dayton Board of Education v. Brinkman, 443 U.S. 526, 536 n. 9, 99 S.Ct. 2971, 2978 n. 9, 61 L.Ed.2d 720 (1979) (Dayton). Nonetheless, “proof of foreseeable consequences is one type of quite relevant evidence of racially discriminatory purpose____” Id. See also Columbus Board of Education v. Penick, 443 U.S. 449, 464-65, 99 S.Ct. 2941, 2949-50, 61 L.Ed.2d 666 (1979) (Columbus ).1 Other circumstantial evidence relevant to the proof of segregative intent includes the historical background and specific sequence of events leading up to the Board’s actions maintaining or exacerbating ethnic imbalance in district schools. See Arlington Heights, 429 U.S. at 267-68, 97 S.Ct. at 564-65. We will focus our review on these latter aspects of the proof, for the defendants concede that they knowingly perpetuated segregation.
B. Standard of Review.
We examine the district court’s findings of fact and its ultimate conclusion as to the Board’s intent under the clearly erroneous standard:
discriminatory intent is a finding of fact to be made by the trial court____ [It] means actual motive; it is not a legal presumption to be drawn from a factual showing of something less than actual motive. Thus, a court of appeals may only reverse a district court’s finding on discriminatory intent if it concludes that the finding is clearly erroneous under Rule 52(a).
Pullman-Standard v. Swint, 456 U.S. 273, 289-90, 102 S.Ct. 1781, 1790-91, 72 L.Ed.2d 66 (1982). Accord Dayton, 443 U.S. at 534, 99 S.Ct. at 2977; Columbus, 443 U.S. at 468, 99 S.Ct. at 2952 (1979) (Burger, C.J., concurring).
Under this standard the district court’s factual finding of lack of segregative intent may be set aside only if, “although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum, 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948); Anti-Monopoly, Inc. v. General Mills Fun Group, Inc., 684 F.2d 1316, 1318 (9th Cir.1982).
In this ease our careful review of the record leaves us with such a firm conviction. The district court has made corn[664]*664prehensive and accurate findings covering the significant historical facts, but we are firmly convinced that it erred when it reached the ultimate conclusion that these facts do not compel a finding of segregative intent.
C. The Neighborhood School Policy.
The significance of the Board’s adherence to a “neighborhood school policy” was central to the district court’s analysis. This policy requires students to attend the schools nearest their homes. The Board has, with few exceptions, adhered to this policy rigidly and refused to allow voluntary transfers for any purpose, including voluntary desegregation.2 Although a neighborhood school policy has been used by school districts to justify segregation, see Dayton, 443 U.S. at 533 n. 7, 99 S.Ct. at 2976 n. 7; Columbus, 443 U.S. at 461-62 & n. 8, 99 S.Ct. at 2948-49 & n. 8; Keyes, 413 U.S. at 206, 93 S.Ct. at 2696, adherence to such a policy is not by itself dispositive of segregative intent. We recognized in our previous opinion that the neighborhood school policy “is merely relevant evidence to be taken into account in deciding whether the forbidden intent did or did not exist.” Diaz II, 612 F.2d at 415. It “does not have constitutional implications one way or the other without a penetrating examination of the complete context in which the neighborhood policy was initially applied and subsequently enforced.” Id.
We must distinguish, however, between a policy requiring students to attend schools within neighborhood attendance areas, and the decision to draw those attendance areas in a manner that achieves or maintains ethnic imbalance. School buildings are large, immobile objects. For many school districts, the location of existing schools will more or less determine feasible attendance areas. Where neighborhoods are themselves racially or ethnically imbalanced, the enforcement of a neighborhood school policy may perpetuate de facto segregation. As we read the Supreme Court’s pronouncements we cannot require a school district with no history of de jure segregation to abandon its neighborhood school policy merely because the infeasibility of relocating its schools has caused the entrenchment of segregation. See, e.g., Swann v. Charlotte-Mecklenburg Board of Education, 402 U.S. 1, 28, 91 S.Ct. 1267, 1282, 28 L.Ed.2d 554 (1971). As we explained in our previous opinion, however, the location and construction of new schools and the closing of old schools are not merely decisions enforcing a neighborhood school policy. Rather, they are decisions that may determine whether the prescribed neighborhood attendance areas will be integrated or segregated. See Diaz II, 612 F.2d at 415. The Supreme Court has recognized that
[i]n the past, choices in this respect have been used as a potent weapon for creating or maintaining a state-segregated school system. In addition to the classic pattern of building schools specifically intended for Negro or white students, school authorities have sometimes, since Brown [v. Board of Education, 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954)], closed schools which appeared likely to become racially mixed through changes in neighborhood residential patterns. This was sometimes accompanied by building new schools in the areas of white suburban expansion farthest from Negro population centers in order to maintain the separation of the races with a minimum departure from the formal principles of “neighborhood zoning”.
[665]*665Swann, 402 U.S. at 21, 91 S.Ct. at 1278. A school district may combine its neighborhood school policy with a segregative pattern of site selection. In such cases, the enforcement of a neighborhood school policy may be one of a series of segregative acts independently evincing segregative intent. Many of these same actions disapproved by the Supreme Court in Swann are also present in this case.
Furthermore, adherence to a neighborhood school policy may have a reciprocal effect by making neighborhoods more segregated.
The construction of new schools and the closing of old ones ... when combined with one technique or another of student assignment, will determine the racial composition of the student body in each school in the system. Over the long run, the consequences of the choices will be far reaching. People gravitate toward school facilities, just as schools are located in response to the needs of people. The location of schools may thus influence the pattern of residential development of a metropolitan area and have important impact on composition of inner-city neighborhoods.
Id. at 20-21, 91 S.Ct. at 1278.
In this case, a series of events unrelated to the ethnic imbalance of the District’s
schools required the Board to make a large number of choices relating to school siting, school construction, the use of portable classrooms and the definition or redefinition of neighborhood attendance areas. Each time, parents, interested organizations and the State of California presented alternative proposals to the Board that could have ameliorated the segregation in the district schools. Most of these opportunities presented no threat to the neighborhood school policy and could have been accomplished within its framework. The Board consistently selected the more segregative alternative.
D. Board Awareness of Segregation and its Duty to Desegregate.
Throughout the period in question the Board was aware of ethnic imbalance in the district schools.3 The Board was also continuously aware that it had an affirmative duty to desegregate under state law.4
In 1963, the California Supreme Court held that:
even in the absence of gerrymandering or other affirmative discriminatory conduct by a school board, a student under some circumstances would be entitled to relief where, by reason of residential seg[666]*666regation, substantial racial imbalance exists in his school____[wjhere such segregation exists, it is not enough for a school board to refrain from affirmative discriminatory conduct. The harmful influence on the children will be reflected and intensified in the classroom if school attendance is determined on a geographic basis without corrective measures. The right to an equal opportunity for education and the harmful consequences of segregation require that school boards take steps, insofar as reasonably feasible, to alleviate racial imbalance in schools regardless of its cause. Our State Board of Education has adopted regulations which encourage transfers to avoid and eliminate racial segregation (Cal.Admin.Code, Title 5, §§ 2010, 2011) ...
Jackson v. Pasadena City School District, 59 Cal.2d 876, 881, 31 Cal.Rptr. 606, 382 P.2d 878 (1963) (emphasis added).
In 1976, the California Supreme Court reapproved the Jackson holding, recognizing that “for more than a decade this court has adhered to the position that school boards in this state bear a constitutional obligation to attempt to alleviate school segregation, regardless of its cause.” Crawford v. Board of Education of the City of Los Angeles, 17 Cal.3d 280, 293, 130 Cal.Rptr. 724, 551 P.2d 28 (1976). The Californiá courts throughout the period consistently had reiterated this obligation of local school boards. See Crawford, 17 Cal.3d at 291-92, 130 Cal.Rptr. 724, 551 P.2d 28 (citing Mulkey v. Reitman, 64 Cal.2d 529, 537, 50 Cal.Rptr. 881, 413 P.2d 825 (1966); San Francisco Unified School District v. Johnson, 3 Cal.3d 937, 957-58, 92 Cal.Rptr. 309, 479 P.2d 669 (1971); Serrano v. Priest, 5 Cal.3d 584, 96 Cal.Rptr. 601, 487 P.2d 1241 (1971)). The district court was clearly erroneous in its finding that the Board’s obligations under state law were in constant flux. See Diaz I, 412 F.Supp. at 334; Diaz III, 518 F.Supp. at 642 (incorporating findings of Diaz I). The obligations were clear and the Board was aware of them.5
Yet, during this period, the Board resisted compliance with state requirements that affirmative action be taken to desegregate its schools. In a 1966 report to the State Department of Education, the Board recognized that it was not meeting the State’s guidelines. The Board acknowledged that it had been subject to “considerable pressure” to alter its neighborhood school policy,6 and stated in the report:
[667]*667District representatives were informed that the State Board of Education felt it had a “mandate” to expedite integration in all schools; that social integration, of all pupils carried the highest priority, regardless; that the traditional neighborhood school concept was no longer valid in districts characterized by “racial imbalance,” and that the subsequent year’s projects (1967-68) would be required to “focus” on intergroup experiences and the general problem of de facto segregation in order to be approved for funding, (emphasis in original).
Despite its recognized duty to take affirmative steps to desegregate, the Board continued to justify actions that maintained segregation by reference to its neighborhood school policy. The Board’s stubborn adherence to this policy in the face of clearly established state law holding desegregation to be a matter of overriding educational importance suggests that the Board was motivated, at least in part, by a desire to avoid desegregation rather than a sincere commitment to the educational benefits of neighborhood schools.7
E. School Siting and Transportation.
Despite the Board’s awareness of the problem of ethnic imbalance and its duty to alleviate it, the Board chose alternatives that perpetuated or intensified segregation and rejected numerous unusual opportunities to reduce segregation that became available throughout the years.
Since 1965, the district has constructed nine new schools on newly-selected sites.8 The Board was aware that state law required it to consider state guidelines on the formation of attendance areas in making siting decisions. Nonetheless, the Board failed to consider the state guidelines9 and [668]*668selected sites with the knowledge that each of the newly constructed schools would be imbalanced from inception, although alternatives existed. Throughout the relevant time period, the Board received a number of communications from state agencies, advising it that district schools were impermissibly segregated and insisting that it follow state guidelines. The Board chose to disregard state law totally.
The Board similarly chose to maintain ethnic imbalance when it closed several northern schools. The district court went through an extensive and careful analysis of the pattern of school closures. See Diaz I, 412 F.Supp. at 319-22, 331-32; Diaz III, 518 F.Supp. at 632-35. The conclusion from this analysis can be summarized easily — little change occurred. Plaintiffs aptly refer to these actions as the “downtown shuffle.” Where northern schools were closed, the Board reassigned students to other schools in the north rather than reassigning them to schools further south as suggested by numerous proposals. The district court noted two particularly disturbing aspects of the decision to close two • downtown junior high schools. First, the restructuring of junior high school attendance areas resulted in two distinct feeder systems, one for the area that was predominantly' Hispanic and another for the area that was predominantly Anglo. Second, the transfer of students transformed one of the district’s few ethnically balanced schools into an imbalanced school. See Diaz III, 518 F.Supp. at 634.
Pursuant to the Field Act,10 the Board demolished thirteen additional schools as unsafe. The Board created a task group to establish a plan for rebuilding the schools. The task group considered several proposals to rebuild the schools in new locations and to redraw attendance areas in order to reduce ethnic imbalance. The district court found that several of the alternatives would have reduced ethnic imbalance. Diaz I, 412 F.Supp. at 331. The Board chose to rebuild most of the Field Act schools on their original sites and to maintain the former attendance areas.11 The Board rebuilt two Field Act schools on newly selected sites. However, it retained the old attendance area boundaries, and both schools remained imbalanced.
During this time the Board considered, and rejected, a proposal to transfer students from the most severely imbalanced Field Act schools to nearby schools in contiguous attendance areas. Washington [669]*669School had a student population of 78.4% Spanish-surnamed. Because of overcrowding, Washington students attended double sessions. One mile from Washington, Riverglen School had a Spanish-surnamed population of only 19.2% and was operating under capacity. The Board refused to redraw attendance areas or to transfer any students from the overcrowded Washington to Riverglen. Students from the Little Orchard neighborhood, which is almost 100% Spanish-surnamed, were assigned for years to Washington even though they actually lived closer to River-glen.
Gardner School had a Spanish-surnamed population of 87.8%. Lincoln Glen, located two miles from Gardner, had a Spanish-surnamed population of 15% and was operating under capacity. Canoas School, located three or four miles from Gardner, had a Spanish-surnamed population of 9.7% and was operating under capacity. The Board never considered transferring students from Gardner to Canoas or to Lincoln Glen.
Despite strong opposition to the rebuilding of Washington and Gardner in their original locations and the presentation of feasible options to decrease segregation, the Board rebuilt them in their original locations and retained the same attendance areas. These actions were taken by the Board in the face of assurances to the State that the Board would use the opportunity presented by the rebuilding of the schools to reduce ethnic imbalance.
School District Superintendent Knight justified the rebuilding of the Field Act schools on the same sites on the ground that it “would not get in the way of a two-way exchange of students.” This justification must be viewed with suspicion in light of the Board’s repeated and vocal opposition to busing to achieve integration. The district court found that “[a]s early as 1963, the board, through resolution, assured the community of its opposition to busing for integration; the board has not deviated from this position.” Id. at 324. The Board proclaimed that it would rebuild on the same sites in order to bus more effectively, while at the same time pronouncing it would never bus because it was seeking other solutions to the segregation problem.
The Board's wide use of busing for everything except integration is itself suspect. In 1973-74,10,431 of 36,000 students were bused daily. All but 14 of the district’s 50 schools used buses to transport students within designated attendance areas. Studies commissioned by the district showed that the total busing required to integrate the district would involve only 6,000 students — 3,000 Hispanic students and 3,000 Anglo students. The Board was clearly not opposed to busing per se, yet it repeatedly voiced its opposition to busing for the purposes of integration. The Board’s use of buses for every purpose but integration, coupled with its intransigent resistance to the use of buses for integration, supports an inference of segregative intent. Cf. Washington v. Seattle School District No. 1, 458 U.S. 457, 102 S.Ct. 3187, 73 L.Ed.2d 896 (1982) (initiative that treated busing for integration differently from busing for other purposes invalid).
In summary, the Board responded to a series of opportunities to select school locations and designate the boundaries of attendance areas by selecting sites and designating boundaries in a manner that ensured continued segregation. “In ascertaining the existence of legally imposed school segregation, the existence of a pattern of school construction and abandonment is ... a factor of great weight.” Swann, 402 U.S. at 21, 91 S.Ct. at 1278-79. See also Dayton, 443 U.S. at 540, 99 S.Ct. at 2980. The Board responded to suggestions for less segregative alternatives by rejecting them. It responded to the guidelines imposed by the State of California by ignoring them.
F. Overcrowding, Faculty Assignment and Educational Priorities.
The Board response to overcrowding was two-fold — the use of portables and double sessions. At one point, the Board instituted double sessions in overcrowded southern [670]*670schools even though some northern schools were operating under capacity.12 The district court found this disturbing because of the known “educationally disadvantageous” effects of double sessions. Diaz III, 518 F.Supp. at 635.
Defendants admit that faculty and staff were intentionally assigned to schools on the basis of race. The Hispanic teachers were concentrated in the downtown schools. The district court found that assignments were not made in order to assign Spanish-speaking teachers to predominantly Mexican-American schools where their language skills might be needed. Diaz III, 518 F.Supp. at 640-41. The district court concluded that the assignment of teachers or staff was not the result of a racially-neutral policy. Id. at 641.
A discriminative pattern of assignment of faculty and staff is among the most important indicia of a segregated school system. Swann, 402 U.S. at 18, 91 S.Ct. at 1277. The district court correctly observed that the law does not permit assignment of faculty on the basis of race even if motivated by a neutral policy, Diaz III, 518 F.Supp. at 640. The district court found the staff and teacher assignments had no effect on student segregation because of the small number of Spanish-surnamed faculty. Its observations may be accurate, but it neither negates the illegality of the assignment policy nor diminishes the inference of racial bias.
The Board rejected an opportunity to reduce segregation through use of portable classrooms. The district owns over 400 portable classrooms and leases others. Portable classrooms, because they are not fixed in one place, provide an inexpensive and flexible way to alter attendance boundaries to reduce overcrowding and ethnic imbalance in particular schools. The district court found that the use of portables to increase integration “was always an alternative available when a need for portables arose; the school district could always have opted to place new portables at the other end of the district from the area of overcrowding____” Diaz III, 518 F.Supp. at 635. Nevertheless, the Board rejected numerous suggestions for portable siting to alter attendance patterns involving transferring Hispanic students from northern schools to schools farther south. The district has never used portables to improve ethnic imbalance.
In one instance in 1974 the Board did depart from its neighborhood school policy when Anglo parents in the southern schools complained about the double sessions. The Board allowed them to go out of the district rather than transfer to the northern schools in which Hispanic students predominated.
At a school board meeting on May 2, 1974, a group of Anglo parents demanded that their children be relieved of double sessions at Muir and Bret Harte junior high schools. On May 16, the board responded by authorizing inter-district transfers permitting Anglo students to attend schools outside their neighborhood attendance areas for the 1974-1975 and 1975-1976 school years. Executive sessions of the board met with the district superintendent and the board attorney to reevaluate the newly-enacted authorization of inter-district transfers. Counsel for the board advised the district that the court in this litigation might adversely interpret deviation from the board’s stated adherence to a neighborhood school policy, in circumstances which relieved crowding while perpetu[671]*671ating ethnic imbalance. On June 20, 1974, after receiving this advice, the board rescinded its transfer resolution.
Diaz I, 412 F.Supp. at 323.
We agree with the advice the Board received from its counsel that the resolution could be interpreted as evidence of segregative intent. The later recission of the resolution because of counsel’s advice does not lessen the implication.
The Board argues that the segregation in district schools results solely from its neighborhood school policy. The evidence in the record, however, demonstrates that the pattern of decisions relating to school siting, school closings, designation of attendance areas, busing, use of portable classrooms, double sessions and transfers maintained or exacerbated ethnic imbalance. None of these decisions was required by the neighborhood school policy; indeed, the Board rejected numerous alternative proposals that were consistent with the neighborhood school policy. The Board’s willingness to depart from its neighborhood school policy in assigning students from the Little Orchard neighborhood to Washington rather than Riverglen and permitting students from the Anglo-dominated Muir and Bret Harte Schools to transfer to other Anglo-dominated schools indicates that the Board was willing to forego its neighborhood school policy in order to maintain segregation. Plaintiffs have presented additional evidence that suggests that the neighborhood school policy itself was a pretext for the Board’s hostility to integration.
G. The Board’s Public Response to Acknowledged Segregation.
The Board’s ostensible attempts to alleviate the problem of school segregation support the conclusion that its adherence to a neighborhood school policy was motivated, at least in part, by segregative intent. The Board’s actions reveal a pattern of resisting suggestions that would have decreased segregation and submitting to political pressure from members of the community opposed to integration. One example is the Board’s response to efforts by the State of California to secure compliance with its integration policy.13 The Board acted in several instances to maintain a posture of commitment to its obligations under state law, although it actually did nothing. In conjunction with the rebuilding of the Field Act schools, the Board assured the State that it would rebuild the schools on different sites to achieve greater ethnic balance. It did not. And, the Board formally adopted the state guidelines on formulation of attendance areas but did not implement them when faced with an opportunity.14 In [672]*672its 1966 report to the State in conjunction with its ethnic survey, the Board cited its 1963 resolution on desegregation and the intent to form a citizen’s committee to study the problem of segregation and recommended solutions. That committee, the Quality Urban Education Action Team (QUEST), was not formed until 1969. QUEST’S history is revealing. The Board repeatedly referred to QUEST in its reports to the state as an indication of progress toward integration, while responding to QUEST in a manner that rendered the committee totally ineffective.
The record shows that at a highly emotional meeting of 1500 persons where “fear and racism” were prevalent, the Board submitted to public pressure and opened QUEST membership to all present — most of whom were opposed to desegregation. See infra note 16. These new members were given their choice of subcommittees on which to serve, and most joined the site location, demographics and magnet school committees. At the same time, the Board modified QUEST’S mandate, instructing it to pay more attention to revision of instructional programs and less attention to attempting to reduce ethnic imbalance without delay.15
Board members told Aaron Harris, Chairperson of QUEST, that the Board would like to dissolve or suspend QUEST because the public perceived that integration would result and that such a perception might have adverse impact upon the upcoming bond election. Over Mr. Harris’s opposition, the executive committee of the reconstituted QUEST voted to suspend activities, a decision in which Harris stated, “the Board happily acquiesced.” Despite these difficulties, QUEST put together a final report making recommendations for solutions to the segregation problem, including magnet schools, education parks, open enrollment, and voluntary busing.16 [673]*673The Board implemented none of QUEST’S suggestions.
Any doubts about whether the Board acted with segregative intent are dispelled by its public statements regarding the bond issue elections. The district court found that these statements
offered 1) the carrot that the bond money would not be used to subsidize busing (and reiterated the Board’s own continued disinclination to use busing for integration purposes) and 2) the stick that failure to pass the bond issue would result in (a) alterations in the neighborhood school policy to spread double sessions throughout the district and (b) the need to send students from the northern end of the district to the south, and/or vice versa.
Diaz III, 518 F.Supp. at 638. The Board catered to a pro-segregation public by mak[674]*674ing assurances that it could not use bond money to finance busing for integration. The message these statements conveyed, and were intended to convey, was unmistakable: if the public did not vote for the bond issue, busing would result. If it did vote for the issue, no busing would result. These statements, considered in the context of the other evidence presented in this case compel an inference of segregative intent.17
III. THE CUMULATIVE WEIGHT OF THE EVIDENCE COMPELS A FINDING OF SEGREGATIVE INTENT.
In making its determination in respect to the Board’s intent, the district court erred in failing to give weight to the cumulative impact of the evidence. See Washington v. Davis, 426 U.S. 229, 242, 96 S.Ct. 2040, 2048, 48 L.Ed.2d 597 (1976); Arlington Heights, 429 U.S. at 266-68, 97 S.Ct. at 563-65; Flores v. Pierce, 617 F.2d 1386, 1389 (9th Cir.), cert. denied, 449 U.S. 875, 101 S.Ct. 218, 66 L.Ed.2d 96 (1980).
The District repeatedly promised the public and state officials that it would desegregate its schools in compliance with state law. Yet, it deliberately ignored state guidelines in making decisions and consistently refused to implement suggestions for desegregation. Similarly, in Columbus, an element of the district court’s finding of segregative intent was the Board’s failure to employ workable suggestions for improving racial balance despite promises otherwise. See Columbus, 443 U.S. at 453, 99 S.Ct. at 2944.18
The Board sited new schools, rebuilt the Field Act schools, used portables and closed schools in a manner that maintained and, in some instances, intensified ethnic imbalance. The Supreme Court found such actions to evince segregative intent in Columbus, 443 U.S. at 462 & n. 11, 467, 99 S.Ct. at 2949 & n. 11, 2951 (site selection and construction),19 in Dayton, 443 U.S. at 532, 533 n. 7, 540, 99 S.Ct. at 2976, 2976 n. 7, 2980 (siting, construction, and addition of classroom space at existing schools), in Swann, 402 U.S. at 20-21, 91 S.Ct. at 1278-1279 (construction and closures); and in Keyes, 413 U.S. at 192, 93 S.Ct. at 2689 (construction, mobile classrooms).
The Board used buses for one-third of its students, but refused to use those buses to achieve integration. See Washington v. Seattle School Dist. No. 1, 458 U.S. 457, 486-87, 102 S.Ct. 3187, 3203-04, 73 L.Ed.2d 896 (1982) (invalidating a Washington initiative that treated busing for integration differently from busing for other purposes). Cf. Swann, 402 U.S. at 18, 91 S.Ct. at 1277 (existing policy of transportation an impor[675]*675tant consideration in identifying a segregated system).
The Board assigned faculty and staff on the basis of ethnic origin without any plausible neutral justification. See Columbus, 443 U.S. at 460, 461, 467, 99 S.Ct. at 2948, 2948, 2951; Dayton, 443 U.S. at 532, 536 n. 9, 99 S.Ct. at 2976, 2978 n. 9; Swann, 402 U.S. at 18-19, 91 S.Ct. at 1277-1278; Green v. County School Board, 391 U.S. 430, 435, 88 S.Ct. 1689, 1692, 20 L.Ed.2d 716 (1968).
The Board responded to overcrowding in the schools by instituting educationally disadvantageous double sessions and departed from its neighborhood school policy to .. , „ . . . , , , . avoid transferring Anglo students to pre- , . ,, XT* • \ , a n 1 dominantly Hispanic schools. See Columbus, 443 U.S. at 461-62 n. 8, n. 9, 99 S.Ct. at 2948-49 n. 8, n. 9 (departure from neighborhood school policy); Dayton, 443 U.S. at 539-40, 99 S.Ct. at 2980-81 (use of optional attendance zones); Keyes, 413 U.S. at 191, 93 S.Ct. at 2688 (manipulation of attendanee zones).
The Board catered to a public that opposed segregation. This is evidenced by the Board’s statements supporting the bond issues, in which it raised the spectre of busing for integration if the levy was not passed. The Board was instrumental in transforming the QUEST committee from a group of concerned citizens to one dominated by those opposed to desegregation. The Board ignored the recommendations the committee made and consistently rejected other suggestions for desegregating its schools.
. . ,, t. j An inescapable conclusion that the Board » . of the evidence as a whole. The pattern of Board choices that consistently maintained or intensified segregation is apparent. Although many of the available alternatives would have presented an incomplete solution, each could have contributed incrementally toward reducing ethnic imbalance. In almost every instance, the Board chose to “turn toward segregation” rather than away from it. Columbus, 443 U.S. at 463 n. 12, 99 S.Ct. at 2949 n. 12. We are left with the firm conviction that the district court was clearly erroneous in its ultimate conclusion that the Board did not act with segregative intent,
jy CONCLUSION
We hold that the San Jose Sch°o1 Board intentionally maintained segregated schools throughout the District. We reniand to the district court for the formation of an appropriate remedy and an award of attorneys’ fees to plaintiffs,
REVERSED and REMANDED,