The Police Officers Association (POA) and three of its members appeal the district court’s approval of the procedures followed for the 1983-84 promotion examinations to Police Sergeant (Q — 50) and Assistant Inspector (Q-35). The POA challenges the district court’s decision to permit the City and County of San Francisco to rescore the promotion examinations in order to comply with its obligations under a Title VII consent decree.
We reverse. We agree with the district court that the examinations as initially weighed were invalid because they had an adverse impact on minorities and women. Rescoring the examinations in order to achieve a specific racial result was, however, improper.
BACKGROUND
The testing resulted from a consent decree entered in consolidated race and sex discrimination actions that the Officers For Justice (OFJ) and the United States brought under 42 U.S.C. §§ 1981 and 1983, and Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e to 2000e-16. The POA intervened in those actions and, with the City, the Civil Service Commission (the Commission), the United States, and the OFJ, agreed to the consent decree. The district court approved the consent decree on March 30, 1979, some six years after the plaintiffs filed the initial complaint. Officers for Justice v. Civil Serv. Comm’n, 473 F.Supp. 801 (N.D.Cal.1979), aff'd, 688 F.2d 615 (9th Cir.1982), cert. denied, 459 U.S. 1217, 103 S.Ct. 1219, 75 L.Ed.2d 456 (1983).
The consent decree required that the City employ good faith efforts to achieve particular goals for employment of women and minorities within specified periods of time. The consent decree prohibited the use of selection procedures that had an adverse impact on women and minorities, unless the City proved the procedures were valid under the Uniform Guidelines on Employee Selection Procedures, 28 C.F.R. 1607. The consent decree specifically prohibited the City.from unlawfully discriminating in any manner on the basis of sex, race, or national origin.
In order to discharge its obligations under the consent decree, the City established a Consent Decree Unit and charged it with developing and administering the promotional examinations. In 1983, the Unit administered the selection procedures for Q-35 Assistant Inspector and Q-50 Sergeant. The promotional examinations had three parts: (1) A multiple choice test to measure technical knowledge and problem solving; (2) a writing skills examination to test the candidates’ written communication skills; and (3) a structured oral examination to measure oral communication, interpersonal qualities, and supervisory abilities. The Commission gave each component of the examination at different times during 1983.
At the time the Unit devised the examination procedures, the Commission did not adopt a rating system. The Consent Decree Division Director and the test creator *1127advised against setting the weights of the components until after the Commission administered the examinations and analyzed the results. Acquiescing to the wishes of the POA, however, the Commission established a weighing procedure part-way through the examination process. On September 12, 1983, after the Commission administered the multiple choice component of the examinations, the Commission established the cut-off scores for the multiple choice component and set the weights for all three components. It set the cut-off scores for the multiple choice examinations at 55 percent for the Q-35 Assistant Inspector and 50 percent for the Q-50 Sergeant. These cut-off scores were above the minimum competency level and did not have an adverse impact on minorities. The Commission set the weights for the examination components as follows:
Q-35 Q-50
Multiple choice: 45% 41%
Written communications 29% ' 29%
Oral examination 26% 30%
After it administered all three components, the Commission informed the parties of the examination results. While it did not reveal the actual identities of the candidates, the Commission ranked them by their assigned examination numbers and specified their race and gender.1 The candidates taking the examinations fell into the following categories:
Total Ethnic Minorities Women
Q-35: 576 148(25.7%) ' 59(10.2%)
Q-50: 640 166 (25.9%) 65 (10.2%)
Based on the September 1983 selection formula, the eligibility statistics were as follows:
Total Ethnic Minorities Women
Q-35: 75 6 (8%) 4 (5.3%)
Q-50: 125 13 (10.4%) 12 (9.6%)
The percentages represented an adverse impact on minorities in both ranks, and a slight adverse impact on women for the Q-35 Assistant Inspector examination. There was no adverse impact on women for the Q-50 Sergeant examination.
The United States and the OFJ objected to the weighing of the test components because the results demonstrated an adverse impact on minorities and women in violation of the consent decree and Title VII. After discovery, the City Attorney recommended that the Commission revise the cut-off and weighing standards. In June 1984, the Commission agreed and revised the scoring procedures for the examinations.
Under the revised formula, the Commission graded the multiple choice and written communication components of the tests on a pass-fail basis, with 55 percent being the passing grade for Q-35 Assistant Inspector and 50 percent being the cut-off grade for Q-50 Sergeant. It set the passing grade for the written communication component at 60 percent for both Q-35 and Q-50. As a result of the cut-off scores, approximately 200 candidates were eliminated.2 The multiple choice examination screened out 25 candidates for Sergeant and 23 for Assistant Inspector. The written examination screened out 130 candidates for Sergeant and 160 for Assistant Inspector. Close to 700 candidates took the examinations. The Commission then ranked the oral examinations of those who survived the multiple choice and written examinations in order. The POA objected to the new selection decision, but the grading proceeded on the basis of the revised weights.
There were 45 vacancies for Q-35 and 75 vacancies for Q-50. Using the revised formula, the top applicants were:
Q-35: 30 Caucasian males
(65.9%)
10 ethnic minorities
(22.29%)
5 women (11.11%)
Q-50: 46 Caucasian males
(61.4%)
18 ethnic minorities
(24%)
11 women (14.6%)
*1128A few days after the Commission adopted the revised formula in June of 1984, the POA filed this action for declaratory and injunctive relief in San Francisco County Superior Court against the City, the Commission and its members. The OFJ later intervened. The defendants removed the case to federal court under the Civil Rights Removal Statute, 28 U.S.C. § 1443(2). Judge Peckham granted a preliminary injunction against the use of the revised formula pending disposition of the action. On March 14, 1985, he granted the City partial summary judgment, holding that Title VII and the consent decree preempted a San Francisco Civil Service rule that required motions for reconsideration to be filed within 30 days of the Commission’s action.3 In May of 1985, Judge Peckham held a fairness hearing to determine if the new formula unnecessarily injured the examination participants.
In June of 1985, Judge Peckham ruled for the City. City and County of San Francisco v. San Francisco Police Officers Association, 621 F.Supp. 1221 (N.D.Cal. 1985). He approved the changes in the selection process and also ruled that the procedures were sufficiently fair. The POA appealed. Both this court and the Supreme Court denied a stay pending appeal. Appointments to the vacancies from the list developed under the revised weights were made in April of 1986 while the appeal was pending.
ANALYSIS
I
Standard of Review
We treat the district court's grant of partial summary judgment as a conclusion of law, which is reviewed de novo. See Foster v. Arcata Assocs., 772 F.2d 1453, 1459 (9th Cir.1985), cert. denied, — U.S. —, 106 S.Ct. 1267, 89 L.Ed.2d 576 (1986). We review the district court’s findings of fact under the clearly erroneous standard. Operating Engineers Pension Trust v. Charles Minor Equip. Rental, Inc., 766 F.2d 1301, 1303 (9th Cir.1985).
II
Mootness
Under Article III, section 2, of the Constitution, the federal courts lack power to decide questions that cannot affect the rights of litigants in the case before them. De Funis v. Odegaard, 416 U.S. 312, 316, 94 S.Ct. 1704, 1705-06, 40 L.Ed.2d 164 (1974) (per curiam). An appeal becomes moot when: (1) it can be said with assurance that there is no reasonable expectation that the alleged violation will recur; and (2) interim relief or events have completely and irrevocably eradicated the effects of the alleged violation. County of Los Angeles v. Davis, 440 U.S. 625, 631, 99 S.Ct. 1379, 1383, 59 L.Ed.2d 642 (1979). The burden of proving mootness is a heavy one. Id.
The City contends that this appeal is moot because the City made the 75 Sergeant and 45 Assistant Inspector appointments while , this appeal was pending and the lists that the POA seeks to invalidate have expired. The City argues that no decision of this court can alter the rights and obligations of the parties.4
The appeal is not moot. The consent decree remains in effect until March of 1989, and the City intends to weigh future promotional examinations in the same manner objected to here. Thus, the violations alleged by the POA may recur. See Sarabia v. Toledo Police Patrolman’s Ass’n, 601 F.2d 914, 916 (6th Cir.1979).
Furthermore, the appointments and the termination of the lists do not remedy the loss of promotional rights of candidates who claim that reweighing discriminated against them. If reweighing were unlawful, the court could order that a new test be devised, which may result in different *1129candidates being promoted than those currently holding the positions.
The City erroneously relies on Western Addition Community Organization v. Alioto, 514 F.2d 542 (9th Cir.), cert. denied, 423 U.S. 1014, 96 S.Ct. 446, 46 L.Ed.2d 385 (1975). The order in Western Addition provided for a temporary hiring arrangement pending the Commission’s execution of an acceptable job-relatedness validation procedure for its written examination. Id. at 544. The issue here concerns normal permanent promotions that may recur, rather than a single emergency situation of a nonrecurring nature. Thus, the appeal is not moot.
Ill
Civil Service Rule 5.07
Section 5.07 of San Francisco’s Civil Service Rules provides that a request for reconsideration of a Commission action must be made within thirty days of the Commission’s initial decision. The rule’s thirty-day time limit is jurisdictional. See Smith v. City and County of San Francisco, 11 Cal.App.3d 606, 610, 89 Cal.Rptr. 878, 879 (1970). The OFJ and the United States challenged the validity of the examinations many months after September 12, 1983, when the Commission set the initial weights. The POA contends that the Commission was therefore time-barred from reconsidering its September 12 decision.
The district court rejected the POA’s argument and found that the Commission properly suspended the application of rule 5.07. Because rule 5.07 would impede the operation of the consent decree and would violate Title VII, the district court determined that the court could have suspended the rule’s application under the Supremacy Clause. The district court further reasoned that the City also had the power to suspend the rule temporarily because courts generally favor voluntary compliance with Title VII by employers.
The POA argues that rule 5.07 is not void under the Supremacy Clause because it does not conflict with Title VII or the consent decree. The POA points out that the rule itself has nothing to do with employment matters, but is rather a statute of limitations governing the jurisdiction of a local body. The POA argues that the rule does not conflict with the consent decree because the Commission could have acted in a timely manner and that it is only the Commission’s lack of diligence that caused the conflict.
Under the Supremacy Clause, any state law that conflicts with federal law or “stands as an obstacle to the accomplishments of the full purposes and objectives of Congress” is void. Lawrence County v. Lead-Deadwood School Dist., 469 U.S. 256, 260, 105 S.Ct. 695, 698, 83 L.Ed.2d 365 (1985). To the extent that a civil service rule conflicts with the operation of a consent decree, it is void under the Supremacy Clause. Cf. Kirkland v. New York State Dept. of Correctional Serv., 711 F.2d 1117, 1132 n. 18 (2d Cir.1983) (court did not consider whether settlement agreement violated state law because state law must yield to federal law in Title VII cases), cert. denied, 465 U.S. 1005, 104 S.Ct. 997, 79 L.Ed.2d 230 (1984); Sarabia, 601 F.2d at 918 (court determined that district court could suspend civil service rule whose application prevented achievement of consent decree goals); United States v. City of Chicago, 549 F.2d 415, 438 (7th Cir.) (insofar as provisions of Illinois Municipal Code as applied violate command of Title VII, their operation must be suspended under Supremacy Clause), cert. denied, 434 U.S. 875, 98 S.Ct. 225, 54 L.Ed.2d 155 (1977). The City properly suspended rule 5.07 because, had the City complied with the rule, the City would have violated its obligations under the consent decree.
The consent decree specifically forbids the City from using tests or other selection criteria that demonstrate an adverse im: pact “unless and until” the City proves the procedures are valid under the Uniform Guidelines. Rather than validate a test that had a proven adverse impact on minorities and women, the City chose to revise its testing procedure. The Uniform Guidelines permit employers to use different se*1130lection procedures that do not have an adverse impact on minorities and women as an alternative to validating challenged test procedures. See 28 C.F.R. § 1607.6A (1986); Adoption of Questions and Answers to Clarify and Provide a Common Interpretation of the Uniform Guidelines on Employee Selection Procedures, 44 Fed.Reg. 11,996, 12,001, Question and Answer n. 31 (1979). Not only does the employer have that option, but here it appears that the Uniform Guidelines required the City to throw out the challenged procedure rather than validate it.
In order to validate a selection procedure under the Uniform Guidelines, the employer must consider alternative selection procedures. 29 C.F.R. § 1607.3B (1986). In examining alternatives, the employer should also examine alternative methods of using the selection procedure. 29 C.F.R. § 1607.5G (1986); 43 Fed.Reg. 38,290,38,291 (1979). For example, an employer may find that a selection procedure that is valid when scored on a pass/fail basis is invalid when used as a ranking device. 29 C.F.R. § 1607.5G (1986). The alternative selection procedure should be used when it has less adverse impact and when the evidence demonstrates that its validity is substantially the same or greater than the challenged procedure. 29 C.F.R. § 1607.3B (1986); 43 FecLReg. 11,996, 12,003-04, Questions and Answers Nos. 50-51 (1979).
The City and the POA agree that the components when reweighed are as job-related as they were under the initial weights. The alternative weighing procedure eliminates, however, the test’s adverse impact. Therefore, the City could not validate the tests under the Uniform Guidelines, as required by the consent decree, because an alternative selection procedure existed that was equally job-related but had no adverse impact.5
In order to comply with its obligations under the consent decree, the City could not use the initial weights. Rule 5.07 stood as a barrier to the City’s ability to revise its testing procedures and, as such, was void under the Supremacy Clause.6
IV
Reweighing Examination Components
The critical issue in this case is whether the Commission acted lawfully in reweighing the examination components. The district court viewed this question in terms of fairness and held a fairness hearing in order to determine if the Commission’s decision to reweigh was a valid affirmative action plan under United Steelworkers of *1131America v. Weber, 443 U.S. 193, 99 S.Ct. 2721, 61 L.Ed.2d 480 (1979).7
In Weber, the Supreme Court identified four criteria that make an affirmative action plan valid under Title VII: (1) it is designed to break down old patterns of racial segregation and hierarchy; (2) it does not create an absolute bar to the advancement of nonminority employees; (3) it is a temporary measure, “not intended to maintain racial balance, but simply to eliminate a manifest racial imbalance”; and (4) it does not unnecessarily trammel the interests of nonminority employees. Weber, 443 U.S. at 208, 99 S.Ct. at 2730. Weber did not hold that these criteria were absolute requirements, but did hold that these aspects of the plan in Weber placed it on the permissible side of the line between permissible and impermissible plans. Id. Here, the district court found that reweighing fit all four Weber criteria and was therefore permissible. We reverse the district court because reweighing the examination unnecessarily trammeled the interests of the nonminority police officers.8
In analyzing whether the interests of nonminorities were unnecessarily trammeled, the district court focused on what rights the candidates possessed and how those rights were affected by reweighing. It determined that the City did not overtly take into account race or sex in the decision to reweigh. It then determined that under California law the only right a person on an eligibility list possesses is the right not to be treated arbitrarily or capriciously. Because the Commission retained all three of the original examination components, using two as screening devices and one as a ranking device, and because the oral component seemed capable of judging the three dimensions it was designed to test, the court determined that the Commission had not acted arbitrarily in reweighing the examinations. The district court was careful to state, however, that its holding was not intended to suggest that the new selection process was the best the City could have chosen to promote candidates. Rather, the district court merely approved of the City’s action under the arbitrary and capricious standard.
.We find that the district court clearly erred when it determined that the decision to reweigh was not a race and gender conscious act. The Commission knew the candidates’ race and gender. It also knew how each candidate performed on the individual components before it chose the oral component as the sole ranking device. Using this information, the Commission res-cored the examination in a manner that achieved a specific racial and gender result. Race and gender were the deciding factors in determining how the test scoring should be restructured. As a race and gender conscious decision, reweighing must be examined under a standard higher than the arbitrary and capricious standard used by the district court in order to protect the interests of nonminorities under Weber.
An affirmative action plan that is carefully contoured to accomplish the limited objective of reducing disparate impact on minority candidates does not unnecessarily trammel the rights of nonminority candidates. See Tangren v. Wackenhut Serv., 658 F.2d 705, 707 (9th Cir.1981), cert. denied, 456 U.S. 916, 102 S.Ct. 1771, 72 L.Ed.2d 175 (1982). “The more closely a plan adheres to its remedial purpose, the *1132less likely it is that the plan will unreasonably infringe upon the interests of other employees.” Johnson v. Transportation Agency, 770 F.2d 752, 758-59 (9th Cir.1984) (citation omitted), aff'd., — U.S. —, 107 S.Ct. 1442, 94 L.Ed.2d 615 (1987). Here, the actions of the Commission could hardly be deemed a carefully contoured affirmative action plan. Rather, reweighing was an ad hoc attempt to achieve a result that the Commission’s promotional examinations failed to produce.
Reweighing unlawfully displaced candidates on the basis of their- race and gender. The information about the candidates’ performance on the individual components led the Commission to choose the oral component as the sole ranking device. If the results of the examinations had been different, the written component or the multiple choice component might have been the new ranking device. Without readministering the test, the Commission examined the results from each component based on race and gender criteria and rescored the test to achieve specific and identified racial and gender percentages. This type of result-oriented scoring is offensive.
Candidates who participate in promotional examinations expect to have an equal opportunity to score well and to achieve promotion. This neutrality cannot exist if the City can rescore the examinations to achieve a particular race and gender balance after it analyzes the results. Permitting an employer to rescore examinations with knowledge of the ultimate results undermines the integrity of the examination process.
Moreover, candidates for promotion should be on notice of how their performance will be evaluated in order to prepare themselves effectively for an examination. Although unintentionally, the Commission misled candidates when it announced the initial weighing system after it administered the multiple choice examination. Of course, candidates attempt to perform well on all sections of an examination. Yet given limited time and resources, a person will concentrate on the parts of the examination that count the most. Here, the emphasis dramatically shifted from an initial scoring structure in which each component was worth approximately one-third of the entire examination, to a weighing system in which the first two components were mere screening devices and a candidate’s rank depended solely on his or her performance on one section of the examination. A test that changes so dramatically after it is administered and the race and gender results are known is no test at all, but is a specious device used to produce whatever result the Commission may choose to achieve.
The plan in Weber did not violate Title VII because it did not deprive white employees of opportunities for advancement, but rather created additional opportunities by establishing new training programs for both blacks and whites. Here, however, the Commission’s decision to reweigh unlawfully restricted the promotional opportunities of nonminority candidates because the tests were scored to achieve a particular racial result. It trammeled the interests of nonminorities, in that the candidates were led to believe that the promotions would be based on merit alone. This harm to nonminorities was unnecessary because a less burdensome alternative, such as administering a new selection procedure, would have better achieved the goals of the consent decree without violating Title VII.
The City was obligated under the consent decree to administer an examination that would not have an adverse effect on minorities and women. When it failed in its first attempt to achieve that goal, the City inappropriately attempted to take short-cuts to meet its obligations. It did so in order to save time. Although we are sympathetic to the City’s time dilemma, using an unlawful procedure is not acceptable. The City was required either to validate its initial examination or, if it could not, to devise and administer an alternative selection procedure that did not have an adverse impact.9
*1133The City was additionally obligated under the consent decree not to practice racial or sexual discrimination — no more against white males than against others. The POA was a party to the consent decree. The POA has a right to insist that this unequivocal renunciation of all discrimination means what it says. The reweighing as practiced here violated the consent decree.
We hold that reweighing the promotion examinations after the results of those examinations were known violated Title VII. Therefore the eligibility lists from which the City made the new appointments for Sergeant and Assistant Inspector are invalid. Because the City was unable to validate its initial weighing system, those results must also be discarded. The City must devise and administer a new selection procedure that complies with this opinion. The scoring of the selection procedure must not depend on the actual results of an examination and must adhere to the requirements of the consent decree and Title VII. We remand to the district court in order that it supervise the development of a proper selection device for promotion.
The judgment is REVERSED and REMANDED.
. Under Civil Service Commission Rules, the City must keep confidential the identity of candidates before a cut-off score is set.
. The 200 number takes into account the fact that some candidates competed for both ranks,
. The Commission adopted the revised formula approximately 8 months after it announced the original weights.
. We grant the City's and the POA’s motions to augment the record on appeal because the developments described in the declarations occurred since the taking of this appeal and are relevant to the panel’s decision on mootness.
. While we find that the Uniform Guidelines permit an employer to choose between validating a selection procedure or developing an alternative selection procedure without an adverse impact, we do not find that the Uniform Guidelines support the Commission’s action in this case. The Uniform Guidelines recognize that an employer may take reasonable steps to assure that members of an underutilized protected group are included in the pool of applicants. See 29 C.F.R. § 1607.17(3) (1986); 43 Fed.Reg. 11,996, 12,001, Question and Answer No. 30 (1979). For the reasons set forth below, however, we find that rescoring, rather than read-ministering the examinations, was unreasonable. The agencies that drafted the guidelines were warned against the use of illegal procedures under the guise of affirmative action. See 43 Fed.Reg. 38,290, 38,293 (1978). The agency response to these concerns was that illegal acts purporting to be affirmative action were not the goal of the guidelines and that employee selection procedures must be lawful. Id. The Commission’s use of the alternative selection procedure was unlawful because it permitted the Commission to manipulate the results of the examinations to produce desired racial and gender percentages. Thus, the Commission's adjustment of scores became the sifting device, rather than the examinations themselves. Administering a new selection device would have achieved the consent decree goals without the appearance of unfairness that resulted from res-coring.
. The POA argues that developing an alternative selection procedure, before attempting to validate the challenged procedure, was an action outside the scope of the consent decree. We disagree. Under the consent decree, which incorporates the Uniform Guidelines, the Commission could choose to revamp its selection procedures rather than seek to validate its initial weights. In fact, the validation process requires that the Commission explore such alternative procedures and use them if they have less adverse impact but are equally job-related. See 29 C.F.R. § 1607.3B (1986).
. The concept of a fairness hearing stems from Bushey v. Civil Serv. Comm'n, 733 F.2d 220, 229 (2nd Cir.1984), cert. denied, 469 U.S. 1117, 105 S.Ct. 803, 83 L.Ed.2d 795 (1985). Bushey concerned the New York State Department of Correctional Services’ written examination for Correction Captain. In that case, the state adjusted minority candidates’ raw test scores in order to eliminate the examination's adverse impact. The district court granted summary judgment in favor of nonminority candidates for Captain and enjoined the state from using the eligibility list based on the adjusted scores. The court of appeals reversed. The court found, however, that the record was insufficient to determine whether the adjustment of scores trammeled the interests of nonminority candidates under Weber. It therefore remanded to the district court for a full exploration of the issue. Id. The hearing on remand in Bushey is the type of proceeding that the district court conducted in this case.
. The POA also challenges the Commission's actions on equal protection grounds. We need not and do not reach the constitutional issue as we find that reweighing violated Title VII.
. To the extent that this opinion conflicts with the result in the Second Circuit case of Kirk*1133land, 711 F.2d at 1133, 1134, we decline to follow it.