Although America began with the vision of a city on a hill, not every American has shared a sense of optimism about our nation's municipalities. Indeed, one of the most illustrious of the Framers regarded great cities as "pestilential to the morals, the health, [and] the liberties of man." Christopher Tunnard, The City of Man 34 (1970) (quoting Thomas Jefferson).

In this vein, American legal institutions have begun over time to view cities with a certain constitutionally based suspicion. Thus, in Monell v. New York City Dep't of Social Servs., 436 U.S. 658, 691, 98 S. Ct. 2018, 2036, 56 L. Ed. 2d 611 (1978), the Supreme Court ruled that municipalities could be held liable under 42 U.S.C. § 1983 for deprivations of federally protected rights which occurred "pursuant to official municipal policy of some nature."1 Monell opened the floodgates for an outpouring of such suits against municipalities.

The case at hand is one example of the genre. At trial, a jury found the City of Fall River (the City) and two municipal officials liable under section 1983 for the passage of a facially neutral ordinance that abolished the plaintiff's job. The defendants' appeals raise a tantalizing question about whether a discriminatory animus displayed by fewer than the minimum number of city council members whose votes would be required to enact an ordinance can (or should) be imputed to the municipality itself. Other interesting questions abound, including questions dealing with causation in the context of constitutional torts and the availability of legislative immunity defenses in that setting. Before addressing any of these issues, however, we must parse Fed.R.App.P. 4(a)(6) for the first time and determine whether the defendants have brought their appeals in a timeous fashion.

Many of the facts in this case are conflicted. We present them as best they have presented themselves, occasionally resolving disparities as the jury permissibly might have done. See, e.g., Veranda Beach Club Ltd. Partnership v. Western Sur. Co., 936 F.2d 1364, 1375 (1st Cir.1991) (discussing standard for appellate review of post-verdict challenges to evidentiary sufficiency).

The City hired the plaintiff, Janet Scott-Harris, as the administrator of the newly created Department of Health and Human Services (HHS). When Scott-Harris entered the City's service in 1987, she became the first African-American ever to hold a managerial position in the municipal government. By all accounts she performed quite well at HHS. Withal, she did not enjoy a problem-free relationship with the City's political hierarchs. In 1988, for example, she clashed with Marilyn Roderick, the vice-president of the City Council. Scott-Harris believed that Roderick made inappropriate references to an aspirant's ethnicity in the course of an employment interview and stormed out of the room. Shortly thereafter, she engaged in a shouting match with Roderick. When Scott-Harris subsequently attempted to apologize, Roderick hung up the telephone.

Scott-Harris' difficulties with Roderick did not end with the aforedescribed incident. There were periodic flare-ups--by way of illustration, Roderick wrote a letter to the City Administrator, Robert Connors, protesting Scott-Harris' use of a City-owned motor vehicle--but it was Scott-Harris' reaction to the dysphemisms spouted by Dorothy (Dot) Biltcliffe, a nutrition program assistant for the City's Council on Aging (COA), that precipitated internecine warfare. In the fall of 1990, Scott-Harris learned that Biltcliffe had been making offensive comments. In one instance, referring to her co-worker Paula Gousie and to Scott-Harris, Biltcliffe remarked: "That little French bitch has her head up that nigger's ass." In another, Biltcliffe referred to a secretary as "a little black bitch." Scott-Harris spoke out against this racist invective and, because COA operated under her general supervision, she consulted with Connors and then drew up a set of charges against Biltcliffe as a prelude to dismissal.

The pendency of these charges did not improve Biltcliffe's manners; she called Scott-Harris "a black nigger bitch" and warned that there would be repercussions because Biltcliffe "knew people." Biltcliffe unabashedly pressed her case with two city councilors (Roderick and Raymond Mitchell) and a state senator who, in turn, called Roderick. After numerous postponements the City held a hearing on March 27, 1991. This resulted in a settlement under which Biltcliffe agreed to accept a 60-day suspension without pay. Mayor Daniel Bogan subsequently intervened and pared the punishment substantially.

During this time frame the City's financial outlook worsened. Municipal officials anticipated that state aid would decline up to 10% in the next fiscal year (July 1, 1991 to June 30, 1992). Mayor Bogan directed Connors to prepare a list of proposed budget cuts to accommodate the anticipated reduction in funding. Connors asked his department heads, including Scott-Harris, for their input. Scott-Harris recommended reducing the hours of school nurses. Bogan rejected this suggestion and, over Connors' objection, insisted that Scott-Harris' position be eliminated.

Because the post had been created by municipal ordinance, its abolition necessitated the same procedural formalities. The City Charter requires the votes of a majority of the nine members of the City Council for passage of such an ordinance. The mayor often submits proposed legislation to the City Council, and, in addition, he must approve every enacted ordinance (or else the Council must override his veto). In February 1991 Bogan asked the Council to do away with Scott-Harris' position. On March 5 the ordinance committee, chaired by Roderick, reported out an emendatory ordinance designed to achieve this end and recommended its passage. Three weeks later the City Council voted six-to-two (Roderick voting with the majority) to approve the position-elimination ordinance. Bogan signed it into law.

At about the same time that he moved to incinerate Scott-Harris' job, Bogan offered her a different portfolio--Public Health Director--which paid approximately $12,000 less per annum. Scott-Harris accepted the offer by letter dated February 28, 1991, but a follow-up communique from Bogan added extra duties and shifted Scott-Harris to a less desirable office. Disappointed, Scott-Harris drafted a letter rejecting the job offer. That letter mysteriously arrived at the mayor's office and was acted upon by Bogan despite Scott-Harris' efforts to retract it. Scott-Harris' tour of duty with the City ended on March 29, 1991--two days after the hearing that led to Biltcliffe's suspension. She filed suit several months later.

Solon, the fabled Greek legislator, once characterized the best type of city as one "in which those who are not wronged, no less than those who are wronged, exert themselves to punish the wrongdoers." Plutarch, Plutarch's Lives 455 (Bernadotte Perrin trans., 1914). Here, the plaintiff's complaint alleged in substance that the City and certain municipal officials2 inverted the Solonic ideal: when the plaintiff responded forcefully (but appropriately) to Biltcliffe's racial slurs, the defendants sided with the wrongdoer and instead punished Scott-Harris by ousting her from her position under a blatant pretext. The plaintiff alleged that, in so doing, the defendants abridged her First Amendment rights and set the stage for redress under section 1983. See Mt. Healthy City Sch. Dist. Bd. of Educ. v. Doyle, 429 U.S. 274, 287, 97 S. Ct. 568, 576, 50 L. Ed. 2d 471 (1977) (explaining that in order to prevail on a section 1983 claim based on the First Amendment, the plaintiff must prove that her protected speech was a substantial or motivating factor in the decision to eliminate her job).

At trial the defendants asserted that their motives in passing the challenged ordinance were exclusively fiscal. The plaintiff disagreed, contending that racial animus and a desire to punish her for protected speech, not budgetary constraints, spurred the introduction and passage of the ordinance. On May 26, 1994, evidently persuaded by the plaintiff's efforts to connect Dot to her dismissal, the jury returned a verdict against all three defendants.3

The verdict form memorialized the jury's conclusions (1) that the plaintiff's constitutionally protected speech was a substantial or motivating factor both in Bogan's decision to recommend enactment of the ordinance and in Roderick's decision to work for its passage, and (2) that these actions proximately caused the extirpation of the HHS director's position. As originally returned, the verdict form added an inconvenient wrinkle; it indicated that the plaintiff had not proven that the City's professed desire to enact the ordinance for budgetary reasons was pretextual. Out of the jury's earshot, the judge expressed her concern that the jury's findings were internally inconsistent. After a brief colloquy, she resubmitted the case to the jury with appropriate supplemental instructions. Shortly thereafter the jury returned a revised verdict form which reiterated everything except the "no pretext" finding. In that wise, the jury, having reconsidered the matter, now concluded that the City's stated reason for wanting the ordinance--budgetary concerns--was not its true reason.

The jury assessed compensatory damages against all three defendants, jointly and severally, in the amount of $156,000; found Bogan liable for punitive damages in the amount of $60,000; and found Roderick liable for punitive damages in the amount of $15,000.4 The court subsequently denied the defendants' motions for judgment notwithstanding the verdict. These appeals followed--but not without a perturbing procedural prelude.

Rule 4(a)(1) of the Federal Rules of Appellate Procedure requires that notices of appeal "be filed with the clerk of the district court within 30 days after the date of entry of the judgment or order appealed from." Compliance with this rule is mandatory and jurisdictional; while a court may construe the rule's strictures liberally, it may not wink at them. See Torres v. Oakland Scavenger Co., 487 U.S. 312, 315, 108 S. Ct. 2405, 2407-08, 101 L. Ed. 2d 285 (1988); Air Line Pilots Ass'n v. Precision Valley Aviation, Inc., 26 F.3d 220, 223 (1st Cir.1994).

In this instance the district court entered the appealable order--the order denying the defendants' post-trial motions for judgment n.o.v.--on January 30, 1995. The defendants did not file their notices of appeal until August of that year. Without more, Rule 4(a)(1) would bar the maintenance of these appeals.

The appeal period denominated by Rule 4(a)(1) is, however, subject to an occasional exception. One such exception, added to the Appellate Rules in 1991, provides:

The district court, if it finds (a) that a party entitled to notice of the entry of a judgment or order did not receive such notice from the clerk or any party within 21 days of its entry and (b) that no party would be prejudiced, may, upon motion filed within 180 days of entry of the judgment or order or within 7 days of receipt of such notice, whichever is earlier, reopen the time for appeal for a period of 14 days from the date of the entry of the order reopening the time for appeal.

Fed.R.App.P. 4(a)(6). The mention of "notice" in Rule 4(a)(6) is a reference to Fed.R.Civ.P. 77(d), which provides:

Immediately upon the entry of an order or judgment the clerk shall serve a notice of entry by mail in the manner provided for in Rule 5 upon each party who is not in default for failure to appear, and shall make a note in the docket of the mailing. Any party may in addition serve a notice of such entry in the manner provided in Rule 5 for the service of papers. Lack of notice of the entry by the clerk does not affect the time to appeal or relieve or authorize the court to relieve a party for failure to appeal within the time allowed, except as permitted in Rule 4(a) of the Federal Rules of Appellate Procedure.

These rules lie at the center of the jurisdictional jumble that confronts us. On the defendants' motions, the district court held a hearing and determined that Fed.R.App.P. 4(a)(6) appropriately could be invoked to excuse the defendants' seeming tardiness. The plaintiff's cross-appeal challenges this determination. Because Rule 4(a)(6) is relatively new, we have not yet had occasion to construe it. We do so today, deciding at the outset that the standard of review which governs a district's court's determinations under Rule 4(a)(6) is abuse of discretion. Accord Nunley v. City of Los Angeles, 52 F.3d 792, 794 (9th Cir.1995).

Certain elements of the Rule 4(a)(6) calculus are essentially undisputed: the defendants were parties entitled to notice of the entry of the appealable final order; their Rule 4(a)(6) motions, filed on April 10 and 11, 1995, came within 180 days of the entry of that order; and no party would be subjected to cognizable prejudice by the granting of the motions. Thus, the decisive questions in this case relate to whether the defendants received notice of the entry of the order within 21 days, and if not, whether they filed their Rule 4(a)(6) motions within seven days of the time when they eventually received such notice.

Both of these questions involve an appreciation of the kind of notice that Rule 4(a)(6) contemplates. In terms, Rule 4(a)(6) advances a unitary concept of notice; its two references to "such notice" plainly relate back to the phrase "notice of the entry of a judgment or order." The problem, exemplified by this case, is that the rule does not specify whether that notice must be written notice or actual notice. That problem defies facile solutions, and the courts of appeals which have addressed it thus far have not achieved consensus. Compare Avolio v. County of Suffolk, 29 F.3d 50, 53 (2d Cir.1994) (holding that the rule contemplates written notice) with Nunley, 52 F.3d at 794 (holding that actual notice suffices) and Zimmer St. Louis, Inc. v. Zimmer Co., 32 F.3d 357, 359 (8th Cir.1994) (same). Though we acknowledge that the phrase, simpliciter, is susceptible of multiple interpretations, we believe that the references to "notice" in Rule 4(a)(6), taken in context, are best read as requiring written notice.

Our starting point is our perception that Appellate Rule 4(a)(6) and Civil Rule 77(d) must be read in pari passu. Accord Nunley, 52 F.3d at 795. The text of Rule 77(d) requires the clerk to serve the notice of entry of an order or judgment "by mail." Because a mailed notice is invariably written, it seems logical to conclude that when reference is made later in the text to "lack of notice of the entry," that reference contemplates lack of written notice.

We think that further evidence to the same effect can be gleaned from the scrivenings of the Advisory Committee. The Advisory Committee's Notes are entitled to weight in interpreting federal rules of practice and procedure. See Whitehouse v. U.S. Dist. Ct. for Dist. of R.I., 53 F.3d 1349, 1364-65 (1st Cir.1995). Here, they tell us that Rule 4(a)(6)provides a limited opportunity for relief in circumstances where the notice of entry of a judgment or order, required to be mailed by the clerk of the district court pursuant to [Rule 77(d) ], is either not received by a party or is received so late as to impair the opportunity to file a timely notice of appeal.

Fed.R.App.P. 4(a)(6), Advisory Committee's Notes. The statement "required to be mailed" refers to "notice of entry of a judgment or order," again suggesting that the notice must be in writing. We believe that when a procedural rule uses the precise phrase employed by the Advisory Committee, it can reasonably be inferred that the phrase means the same thing in both contexts.

Policy concerns point us in the same direction. Reading Rule 4(a)(6) to require written notice will simplify future proceedings. As the familiar request to "put it in writing" suggests, writings are more readily susceptible to proof than oral communications. In particular, the receipt of written notice (or its absence) should be more easily demonstrable than attempting to discern whether (and, if so, when) a party received actual notice. Such a scheme not only takes much of the guesswork out of the equation, but also, because Rule 77(d) specifically provides that parties who do not wish to rely upon the clerk to transmit the requisite written notice may do so themselves, the scheme confers certitude without leaving a victorious litigant at the mercy of a slipshod clerk.

To sum up, we hold that written notice is required to trigger the relevant time period under Rule 4(a)(6); oral communications or other forms of actual notice will not serve. We now apply this holding to the facts at hand.

The district court found that the defendants did not receive written notice of the entry of the operative order until April 7, 1995, when the plaintiff's counsel sent them a demand letter seeking satisfaction of the judgments. The court made this finding against a backdrop of unusual events. The defendants' motions for judgment n.o.v. were argued on September 29, 1994. During that session, an unrecorded sidebar conference occurred. The court's comments at that conference left all counsel with the distinct impression that an appealable final judgment would not enter until the court decided the plaintiff's pending application for attorneys' fees. Although the impression was mistaken, see Budinich v. Becton Dickinson & Co., 486 U.S. 196, 202-03, 108 S. Ct. 1717, 1721-22, 100 L. Ed. 2d 178 (1988) (holding that the appeal period commences once a final decision on the merits has been entered, irrespective of any claim for attorneys' fees), it proved persistent. The plaintiff's lawyer, no less than defense counsel, labored under the misimpression; he wrote to the defense team on February 2, 1995, stating in relevant part: "I received the Court's memorandum and order on the defendants' motion for J.N.O.V. The only remaining issue before judgment can be entered is the plaintiff's unopposed motion for attorney's fees." (Emphasis supplied).

Unbeknownst to the parties, however, the court had granted the plaintiff's motion for attorneys' fees in late 1994. The clerk entered this order on the docket but, apparently, neglected to serve copies of the order or the docket entry on counsel. To complicate matters further, when defense counsel made inquiries to the clerk in February and March of 1995 as to whether an order had been entered disposing of the fee application, the clerk said that one had not.

Last but not least, although all counsel in one way or another had actual notice of the order that denied the defendants' motions for judgment n.o.v. by February 1, 1995, cases discussing Rule 4(a)(6) differentiate between notice of an order and notice of the entry of the order, indicating that the rule contemplates the latter. See Virella-Nieves v. Briggs & Stratton Corp., 53 F.3d 451, 452-54 (1st Cir.1995). In this instance the clerk attempted to furnish such notice, but one copy of the court's order was addressed incorrectly and returned by the Post Office as undeliverable, while another copy, plucked by a different lawyer from the clerk's office, bore no notation that it had been entered on the docket. From this tangled record the district court concluded that, though at least one defense attorney received actual notice of the entry of the order on February 24, 1995,5 it was not until April 7, 1995--when the plaintiff's attorney demanded satisfaction of the judgments--that the defendants received a written notice sufficient to animate Rule 4(a)(6). They filed their excusatory motions within seven days of their receipt of this notice.

Given these facts, and given the confused circumstances that contributed to the muddle, the district court did not abuse its discretion in finding that the requirements of Rule 4(a)(6) had been met and in reopening the time for appeal. Since the defendants all filed their notices of appeal within the 14-day period that began on August 14, 1995, when Judge Saris entered her order reopening the time for doing so, we conclude that the appeals are properly before us.

The defendants collectively assert that the district court erred in refusing to declare a mistrial when presented with the original verdict form and added impudence to injury by resubmitting the case for further deliberation. We review the district court's denial of the defendants' motions for a mistrial for abuse of discretion. See Clemente v. Carnicon-P.R. Mgmt. Assocs., 52 F.3d 383, 388 (1st Cir.1995). We evaluate the judge's related actions, namely, her decisions to reject the original verdict form and to resubmit the matter, under the same standard of review. See Santiago-Negron v. Castro-Davila, 865 F.2d 431, 444 (1st Cir.1989).

The defendants' argument on this point boils down to a claim that the district court crafted a verdict form that was structurally flawed; that the jury responded to it by returning two irreconcilable findings; and that, therefore, Judge Saris should have granted the defendants' motions for a mistrial. But it is not enough to preserve the defendants' point that, after the jury first returned with the verdict form, the defendants pounced on the perceived inconsistency and moved to pass the case. Rather, the viability of this assignment of error harks back to the circumstances surrounding the emergence of the verdict form. Although the defendants now say that the form tempted potential confusion, they failed to object when the judge initially submitted it to the jury. The failure to object to the structure of a verdict form before the jury retires, like the failure to object to any other portion of the judge's charge, constitutes a waiver. See Fed.R.Civ.P. 51; see also Phav v. Trueblood, Inc., 915 F.2d 764, 769 (1st Cir.1990) (holding that Rule 51 applies to verdict forms as well as to the trial court's oral instructions); Anderson v. Cryovac, Inc., 862 F.2d 910, 918 (1st Cir.1988) ("If a slip has been made, the parties detrimentally affected must act expeditiously to cure it, not lie in wait and ask for another trial when matters turn out not to their liking.").

We need not probe this point too profoundly, for in all events the judge handled the perceived incongruity in an agreeable manner. When a verdict appears to be internally inconsistent, the safest course--in the absence of irreparable damage, and none appears here--is to defer its acceptance, consult with counsel, give the jury supplemental instructions, and recommit the matter for further consideration. See Hafner v. Brown, 983 F.2d 570, 575 (4th Cir.1992) ("If the district judge concludes that an inconsistent verdict reflects jury confusion or uncertainty, he or she has the duty to clarify the law governing the case and resubmit the verdict for a jury decision."); Poduska v. Ward, 895 F.2d 854, 856 (1st Cir.1990) (deeming it "precisely correct" for a judge, faced with an unclear and inconsistent jury verdict, to provide supplemental instructions and then recommit the matter to the jury). This is exactly the course of action that Judge Saris followed. The actual instructions that she gave, first orally and then in a written response to a jury question, were unimpugnable.6 We discern no error, no unfairness, and no abuse of discretion either in the judge's handling of matters related to the verdict form or in her denial of the defendants' motions for a mistrial.

We turn now to the City's principal assignment of error. Clearly, a municipality may be held liable under section 1983 for the passage of a single ordinance or piece of legislation. See, e.g., Pembaur v. City of Cincinnati, 475 U.S. 469, 480, 106 S. Ct. 1292, 1298-99, 89 L. Ed. 2d 452 (1986). Although municipal liability cannot be based on the doctrine of respondeat superior in this context, see Monell, 436 U.S. at 691, 98 S.Ct. at 2036, such liability can flow from a finding that the city itself has acted through an official decision of its legislative body.7 Hence, from a purely theoretical standpoint, nothing prevents a determination that, if the ordinance here in question--which was passed by a majority vote of the Fall River City Council and approved by the mayor--violates the plaintiff's First Amendment rights, then the City is liable for the violation under section 1983.

We pause at this juncture. We think it is important to note early on that the defendants have not challenged the premise, or the district judge's confirmatory ruling, that Scott-Harris' speech was protected by the First Amendment in the sense needed to give rise to a claim under section 1983. Yet the Supreme Court has laid down important restrictions: to give rise to a section 1983 action, a plaintiff's speech must have been on a matter of public concern, and her interest in expressing herself must not be outweighed by the state's interest as employer in promoting the efficiency of the services that it performs. See Waters v. Churchill, 511 U.S. 661, 667-69, 114 S. Ct. 1878, 1884, 128 L. Ed. 2d 686 (1994); Connick v. Myers, 461 U.S. 138, 142, 103 S. Ct. 1684, 1687, 75 L. Ed. 2d 708 (1983).

Given the Supreme Court's application of these tests in Connick, 461 U.S. at 147-54, 103 S.Ct. at 1690-94, one could argue that Scott-Harris' comments about, and efforts to discipline, a particular employee do not qualify as speech on a matter of public concern. We do not pursue this point because it has not been argued to us; it has, therefore, effectively been waived. We mention it, however, because we do not intend our opinion to be taken as deciding that the facts here asserted comprise protected speech.

We note, moreover, that there is another unusual twist to this case. In most similar instances, the constitutional deprivation is apparent on the face of the ordinance or in the text of the challenged municipal policy, thus eliminating any need for a predicate inquiry into the motives of individual legislators. See, e.g., City of Oklahoma City v. Tuttle, 471 U.S. 808, 822-23, 105 S. Ct. 2427, 2435-36, 85 L. Ed. 2d 791 (1985); City of Newport v. Fact Concerts, Inc., 453 U.S. 247, 251-53, 101 S. Ct. 2748, 2751-53, 69 L. Ed. 2d 616 (1981); Bateson v. Geisse, 857 F.2d 1300, 1303 (9th Cir.1988); Little v. City of N. Miami, 805 F.2d 962, 967 (11th Cir.1986); 18A James Perkowitz-Solheim et al., McQuillin Mun. Corp. § 53.173 (3d ed. 1993). Here, by contrast, the City enacted an ordinance which, on its face, is benign. In cases like this one, implicating the exercise of First Amendment rights, liability under section 1983 can attach to the passage of a facially benign law only if one peers beneath the textual facade and concludes that the legislative body acted out of a constitutionally impermissible motive. This is a delicate business, but this court previously has sanctioned an investigation into the motives that underlay the enactment of a facially neutral ordinance for the purpose of assessing liability under section 1983, see Acevedo-Cordero v. Cordero-Santiago, 958 F.2d 20, 23 (1st Cir.1992), and we are bound by that precedent.

Still, the accumulated jurisprudence leaves perplexing problems of proof unanswered. The baseline principle is well-settled: legislators' bad motives may be proven by either direct or circumstantial evidence. See, e.g., United States v. City of Birmingham, 727 F.2d 560, 564-65 (6th Cir.), cert. denied, 469 U.S. 821, 105 S. Ct. 95, 83 L. Ed. 2d 41 (1984); Smith v. Town of Clarkton, 682 F.2d 1055, 1064-65 (4th Cir.1982). But this principle speaks to the qualitative nature of the evidence that is gathered; it does not address the quantitative question. That question is best framed as follows: How many municipal legislators (or, put another way, what percentage of the legislative body) must be spurred by a constitutionally impermissible motive before the municipality itself may be held liable under section 1983 for the adoption of a facially neutral policy or ordinance? This is a difficult question, and the case law proves a fickle companion.

Some courts appear to have held that the plaintiff must adduce evidence sufficient to show that a majority of the members of the legislative body acted from a constitutionally proscribed motive before this kind of municipal liability can attach. Often this position is implied rather than specifically articulated. See generally United States v. City of Yonkers, 856 F.2d 444, 457-58 (2d Cir.1988). But some courts have been more forthcoming. In Church v. City of Huntsville, 30 F.3d 1332 (11th Cir.1994), a group of homeless persons alleged that the city had adopted a policy of excluding them from the community. The plaintiffs based their section 1983 action on the acts and statements of one individual on a five-member city council. The court observed that a single council member did not have any authority either to establish municipal policy or to bind the municipality. See id. at 1343-44. It therefore examined the evidence against the other four councilors, finding that two had opposed the alleged policy and that two had expressed no views on the subject. The court refused to draw an inference of discriminatory intent from the silence of council members, see id. at 1344 n. 5, and rejected the plaintiffs' claim.

Other courts, acting principally in the areas of race and gender discrimination, have not required evidence of the motives of a majority of the legislative body before imposing liability on the municipality under section 1983. Representative of this line of cases is United States v. City of Birmingham, 538 F. Supp. 819 (E.D.Mich.1982), aff'd, 727 F.2d 560 (6th Cir.1984). There, the district court held a city liable for violations of the Fair Housing Act, 42 U.S.C. §§ 3604(a), 3617 (1994), based on the actions of a seven-member municipal commission which had blocked the construction of racially-integrated housing by a four-to-three vote. While opponents of the project had attributed their position to a series of articulated nondiscriminatory rationales, the court looked behind their avowals and ruled, based on a combination of direct and circumstantial evidence, that racial considerations actually propelled the commission's action. 538 F. Supp. at 826-27. The court concluded that the city could be held liable for the commissioners' animus even though there was no proof of the motives of all four commissioners who voted to kill the project; it was enough, the court suggested, if "racial considerations were a motivating factor among a significant percentage of those who were responsible for the city's [rejection of the project]." Id. at 828. Explicating this construct, the court indicated that a "significant percentage" would not have to encompass the entire four-person majority. See id. at 828-29. Noting evidence that racial concerns motivated "at least two of the four members of the majority faction," the court declared that "[t]hat fact alone may be sufficient to attribute a racially discriminatory intent to the City." Id. at 829.8

Two Massachusetts cases also premise municipal liability on evidence concerning less than a majority of the relevant legislative body. In Southern Worcester County Regional Voc. Sch. Dist. v. Labor Relations Comm'n, 386 Mass. 414, 436 N.E.2d 380 (1982), the Supreme Judicial Court (SJC) upheld a lower court's finding that the plaintiffs had been discharged based on their union activity. The SJC declared that "it is not fatal to the [plaintiffs'] claims that only three of the seven members of the school committee made anti-union statements." Id. 436 N.E.2d at 385. The court concluded that the three members' statements, coupled with evidence of bias on the part of the school superintendent (who had no vote), sufficed to support the finding of liability. See id. Similarly, in Northeast Metro. Regional Voc. Sch. Dist. Sch. Comm. v. MCAD, 31 Mass.App.Ct. 84, 575 N.E.2d 77 (1991), a gender discrimination case involving a refusal to hire, the court noted that direct evidence of bias had been exhibited by only two of the twelve members of the school committee. See id. 575 N.E.2d at 81. The court upheld a finding of liability based on this evidence and on statements by three other committee members that the plaintiff had been a victim of discrimination and/or had been the best qualified candidate for the job. See id. at 81-82.

The precedent in this area is uncertain, and persuasive arguments can be made on both sides. On the one hand, because a municipal ordinance can become law only by a majority vote of the city council, there is a certain incongruity in allowing fewer than a majority of the council members to subject the city to liability under section 1983. On the other hand, because discriminatory animus is insidious and a clever pretext can be hard to unmask, the law sometimes constructs procedural devices to ease a victim's burden of proof. See, e.g., McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-05, 93 S. Ct. 1817, 1824-26, 36 L. Ed. 2d 668 (1973) (establishing presumptions for use in Title VII cases); Mesnick v. General Elec. Co., 950 F.2d 816, 823-24 (1st Cir.1991) (adopting comparable format for age discrimination cases), cert. denied, 504 U.S. 985, 112 S. Ct. 2965, 119 L. Ed. 2d 586 (1992). Where, as here, a plaintiff alleges that a city's councilors connived to victimize her by the pretextual passage of a facially neutral ordinance, it may be overly mechanistic to hold her to strict proof of the subjective intentions of a numerical majority of council members.

Cognizant of these competing concerns, we eschew for the time being a bright-line rule. Rather, we assume for argument's sake (but do not decide) that in a sufficiently compelling case the requirement that the plaintiff prove bad motive on the part of a majority of the members of the legislative body might be relaxed and a proxy accepted instead. Nevertheless, any such relaxation would be contingent on the plaintiff mustering evidence of both (a) bad motive on the part of at least a significant bloc of legislators, and (b) circumstances suggesting the probable complicity of others. By way of illustration, evidence of procedural anomalies, acquiesced in by a majority of the legislative body, may support such an inference. See, e.g., City of Birmingham, 727 F.2d at 564-65; Town of Clarkton, 682 F.2d at 1066-67. By like token, evidence indicating that the legislators bowed to an impermissible community animus, most commonly manifested by an unusual level of constituent pressure, may warrant such an inference. See, e.g., United States v. Yonkers Bd. of Educ., 837 F.2d 1181, 1221-25 (2d Cir.1987), cert. denied, 486 U.S. 1055, 108 S. Ct. 2821, 100 L. Ed. 2d 922 (1988); City of Birmingham, 538 F. Supp. at 824-27. The key is likelihood: Has the plaintiff proffered evidence, direct or circumstantial, which, when reasonable inferences are drawn in her favor, makes it appear more probable (i.e., more likely than not) that discrimination was the real reason underlying the enactment of the ordinance or the adoption of the policy?

The facts of this case do not require that we refine the point to any further extent. Scott-Harris has not only failed to prove that a majority of the councilors possessed a bad motive, but she also has failed to furnish enough circumstantial evidence to ground a finding that, more likely than not, a discriminatory animus propelled the City Council's action.

The evidence, viewed most hospitably to the plaintiff,9 reveals that six of the nine councilors voted in favor of the challenged ordinance and two opposed it. The plaintiff presented sufficient evidence from which a jury could deduce that one of these six, Roderick, along with Mayor Bogan (who did not have a vote), acted out of a bad motive.10 The plaintiff also produced some glancing evidence apropos of Councilor Mitchell: he and Roderick were friends; Roderick spoke to him about the Biltcliffe/Scott-Harris imbroglio; and Biltcliffe called him, presumably to protest her treatment. The jury could have found from other evidence in the case that Mitchell probably voted in favor of the ordinance (although the record does not eliminate the possibility that he abstained). Even though Mitchell did not testify and the substance of his conversations with Roderick and Biltcliffe are unknown, we assume arguendo that a jury reasonably could infer that Mitchell, too, acted for a proscribed reason.

The remaining gaps in the plaintiff's proof are considerably more difficult to overlook. None of the other seven city council members uttered any untoward statements or engaged in any suspicious actions. The "we must slash the budget" pretext had a ring of plausibility, and from aught that appears, none of these seven individuals had any way of knowing that the position-elimination ordinance would not save the City sorely needed funds. Nor is there strong circumstantial evidence of complicity; indeed, the record tells us almost nothing about the inclinations of the silent seven.11 Moreover, the plaintiff made virtually no effort to adduce such evidence. She neither deposed any of the seven nor called them as witnesses at trial. She did not attempt to show that any of the other four councilors who voted for the ordinance had any basis for doubting the truth of the party line ("we must slash the budget") or that they possessed ties to Roderick or Bogan, or that they were beholden to Biltcliffe, or that they were hostile to Scott-Harris. The stark fact is that the motivations of the council members--other than Roderick and Mitchell--did not receive individualized scrutiny. By any responsible standard, this sparse evidence falls short of providing a proper predicate for a finding of municipal liability.

We do not think it is a coincidence that in every analogous case in which municipal liability has been imposed on evidence implicating less than a majority of a legislative body, substantial circumstantial evidence existed from which the requisite discriminatory animus could be inferred. In City of Birmingham, the evidence showed that the race-based opposition of constituents to integrated housing was widespread, pronounced, and vociferously articulated. After several members who supported the racially integrated development were ousted from office, the commission responded to this unremitting pressure and took the unprecedented step of submitting the proposal to a community referendum. 538 F. Supp. at 826-29. In Yonkers Bd. of Educ., the requisite inference was supported by evidence of massive constituent agitation as well as by "departures from the normal procedural sequence" in respect to the challenged proposal. 837 F.2d at 1221.

In this case no such evidence exists. Nothing suggests the City Council deviated from its standard protocol when it received and enacted the ordinance that abolished the plaintiff's job. Nothing suggests that the vote took place in an atmosphere permeated by widespread constituent pressure.12 Putting speculation and surmise to one side, it simply cannot be inferred that more than two of the council members who voted to abolish the plaintiff's position did so to punish her for protected speech. We cannot rest municipal liability on so frail a foundation. Because no reasonable jury could find against the City on the proof presented, Fall River's motion for judgment as a matter of law should have been granted.

Roderick and Bogan advance a different constellation of arguments in support of their motions for judgment n.o.v. We treat these arguments sequentially.

A. Legislative Immunity.

The individual defendants concentrate most of their fire on the district court's rendition of the doctrine of legislative immunity. While municipalities do not enjoy immunity from suit under section 1983, see Leatherman v. Tarrant County Narcotics Intell. & Coord. Unit, 507 U.S. 163, 166, 113 S. Ct. 1160, 1162, 122 L. Ed. 2d 517 (1993), lawmakers have absolute immunity from civil liability for damages arising out of their performance of legitimate legislative activities. See Tenney v. Brandhove, 341 U.S. 367, 376, 71 S. Ct. 783, 788, 95 L. Ed. 1019 (1951); National Ass'n of Social Workers v. Harwood, 69 F.3d 622, 629-30 (1st Cir.1995). This immunity derives from federal common law and, under existing Supreme Court precedents, embraces state lawmakers, see Tenney, 341 U.S. at 376, 71 S.Ct. at 788, and regional officials, see Lake Country Estates, Inc. v. Tahoe Regional Planning Agency, 440 U.S. 391, 405, 99 S. Ct. 1171, 1179, 59 L. Ed. 2d 401 (1979).13

The Court has yet to decide whether local legislators are protected by this strain of absolute immunity, see Lake Country Estates, 440 U.S. at 404 n. 26, 99 S. Ct. at 1178 n. 6 (reserving the question), but the lower federal courts, including this court, have shown no reticence in holding that the doctrine of legislative immunity is available to such persons. See, e.g., Acevedo-Cordero, 958 F.2d at 22-23; Aitchison v. Raffiani, 708 F.2d 96, 98-100 (3d Cir.1983); Reed v. Village of Shorewood, 704 F.2d 943, 952-53 (7th Cir.1983); Bruce v. Riddle, 631 F.2d 272, 274-80 (4th Cir.1980). We reaffirm today that the shield of legislative immunity lies within reach of city officials.

This holding does not end our inquiry. Although legislative immunity is absolute within certain limits, legislators are not immune with respect to all actions that they take. The dividing line is drawn along a functional axis that distinguishes between legislative and administrative acts. The former are protected, the latter are not. See Acevedo-Cordero, 958 F.2d at 23. We have used a pair of tests for separating the two:

The first test focuses on the nature of the facts used to reach the given decision. If the underlying facts on which the decision is based are "legislative facts," such as "generalizations concerning a policy or state of affairs," then the decision is legislative. If the facts used in the decision making are more specific, such as those that relate to particular individuals or situations, then the decision is administrative. The second test focuses on "the particularity of the impact of the state of action." If the action involves establishment of a general policy, it is legislative; if the action "single[s] out specifiable individuals and affect[s] them differently from others," it is administrative.

Cutting v. Muzzey, 724 F.2d 259, 261 (1st Cir.1984) (citations omitted).

When the relevant facts are uncontroverted and sufficiently developed, the question whether an act is "administrative" as opposed to "legislative" is a question of law, and it may be decided by the judge on a pretrial motion. See Acevedo-Cordero, 958 F.2d at 23. When the material facts are genuinely disputed, however, the question is properly treated as a question of fact, and its disposition must await the trial. See id.

In some ways, Acevedo-Cordero and this case are fair congeners. There, as here, the defendants asserted that budgetary woes sparked the enactment of a facially benign position-elimination ordinance. There, as here, the plaintiff(s) countered with a charge that, in fact, a constitutionally proscribed reason lurked beneath the surface. There, as here, conflicted evidence as to the defendants' true motives raised genuine issues of material fact. Acevedo-Cordero teaches that in such situations the issue of immunity must be reserved for the trial. See id.

Judge Saris faithfully applied these teachings, refusing to reward premature attempts by the individual defendants to dismiss the action on the basis of legislative immunity. At the end of the trial, the jury made two crucial findings. First, it found that the defendants' stated reason for enacting the position-elimination ordinance was not their real reason. Second, it found that the plaintiff's constitutionally sheltered speech was a substantial or motivating factor in the actions which Roderick and Bogan took vis-a-vis the ordinance. These findings reflect the jury's belief that the individual defendants relied on facts relating to a particular individual--Scott-Harris--in the decisionmaking calculus and devised an ordinance that targeted Scott-Harris and treated her differently from other managers employed by the City.

We think that in passing on the individual defendants' post-trial motions, the judge in effect accepted these findings and concluded that the position-elimination ordinance (which, after all, constituted no more in this case than the means employed by Scott-Harris' antagonists to fire her) constituted an administrative rather than a legislative act. As long as the quantum of proof suffices--a matter to which we shall return--both this conclusion and its natural corollary (that Roderick and Bogan are not shielded from liability by operation of the doctrine of legislative immunity) rest on solid legal ground.14 See, e.g., Negron-Gaztambide v. Hernandez-Torres, 35 F.3d 25, 27-28 (1st Cir.1994), cert. denied, 513 U.S. 1149, 115 S. Ct. 1098, 130 L. Ed. 2d 1066 (1995); Vacca v. Barletta, 933 F.2d 31, 33 (1st Cir.), cert. denied, 502 U.S. 866, 112 S. Ct. 194, 116 L. Ed. 2d 154 (1991).

B. Causation.

Roderick has another string to her bow. She posits that, as a matter of law, her actions in respect to the position-elimination ordinance cannot be deemed the proximate cause of the harm to Scott-Harris.15 She bases this claim on the fact that her vote alone was impuissant: five votes would ensure enactment of the ordinance, but six legislators voted for passage. Thus, not only was she unable to get the ordinance enacted by herself, but it also would have been passed without her cooperation. This thesis has a patina of plausibility, but it misstates the question before us (and, consequently, we take no view of it).

According to accepted lore, section 1983 actions are to be considered against the background of traditional tort principles. See Monroe v. Pape, 365 U.S. 167, 187, 81 S. Ct. 473, 484, 5 L. Ed. 2d 492 (1961); Wagenmann v. Adams, 829 F.2d 196, 212 (1st Cir.1987). In tort law, determinations relating to causation are customarily "question[s] of fact for the jury, to be solved by the exercise of good common sense in the consideration of the evidence of each particular case." Springer v. Seaman, 821 F.2d 871, 876 (1st Cir.1987) (citations omitted). Phrased another way, "[a]pplication of the legal cause standard to the circumstances of a particular case is a function ordinarily performed by, and peculiarly within the competence of, the factfinder." Swift v. United States, 866 F.2d 507, 510 (1st Cir.1989).

In this instance, the judge charged the jury as follows:

The defendant's actions are the legal cause of the plaintiff's injuries if [they were] a substantial factor in bringing about the harm.... It does not matter whether other concurrent causes contributed to the plaintiff's injuries so long as you find that the defendant's actions were a substantial factor in producing them. If defendant's actions were a substantial factor, then they were the legal cause or what we call the proximate cause.

Because no one objected to these instructions, they, whether or not entirely accurate, are the law of the case.16 See Moore v. Murphy, 47 F.3d 8, 11 (1st Cir.1995); Milone v. Moceri Family, Inc., 847 F.2d 35, 38-39 (1st Cir.1988).

We believe that the jury, applying this standard to the facts before it, could reasonably have concluded that Roderick's overall conduct was a substantial factor in depriving the plaintiff of her constitutional rights. After all, Roderick was not just another face in the crowd: she served as vice-president of the City Council and chaired its ordinance committee; as a result, the jury easily could find that she played a role in the passage of the ordinance that was disproportionate to her single vote. In order to gain approval, the ordinance had to go through the five-member ordinance committee. Roderick established this committee's agenda, and its favorable report on March 5 cleared the way for the ordinance's enactment.17

Although the plaintiff's evidence in this regard is not robust, it suffices in the context of the record as a whole to render the issue of causation susceptible to differing evaluative determinations. Thus, the district judge did not err in submitting the causation question to the jury. And because the jury reasonably could have adopted one such view of the evidence and concluded that Roderick made a successful effort to have the plaintiff ousted, the liability finding must stand.

C. Sufficiency of the Evidence.

Roderick and Bogan, in chorus, assert that insufficient evidence exists from which a jury lawfully could find that the desire to punish the plaintiff for her protected speech was a substantial or motivating factor behind the actions which they took. This assertion is easily refuted.

In challenging a jury verdict on sufficiency grounds, a defendant labors under a heavy burden. See supra note 9 (elucidating applicable legal standard and citing cases). Because the evidence in this case is capable of supporting two sets of divergent inferences, Roderick and Bogan cannot carry their burden.

We choose not to tarry. It suffices to say that, on this pleochroic record, the jury could have found that Biltcliffe used political connections to hinder the investigation of Scott-Harris' accusations by, inter alia, banishing the accuser, and that Roderick and Bogan were the instruments of her vengeance. Roderick bore an animosity toward Scott-Harris based on a history of friction between the two women, and the jury permissibly could have found that when Biltcliffe complained to her about Scott-Harris' charges, she spoke to Connors; that when Scott-Harris persisted, Roderick agreed to push the position-elimination ordinance despite the fact that Scott-Harris was performing her duties well; that the asserted budgetary basis for the ordinance was a sham;18 and that Roderick knew as much.

As to Bogan, much of the same evidence is relevant. In addition, the jury could have found that he knew Biltcliffe and resented Scott-Harris' outspoken efforts to cashier her; that he abetted the effort to save Biltcliffe's sinecure by terminating Scott-Harris (and no other manager) for a bogus reason; that he proposed the position-elimination ordinance to that end, notwithstanding Connors' opposition; that he happily signed it into law; that when he learned of Scott-Harris' intention to accept a different municipal position at a reduced salary, he pulled the rug from under her by increasing the responsibilities of the job and shifting her to a dingy office; that when Scott-Harris tried to retract her rejection of this diminished position, he foiled her efforts to do so; and that in all events Bogan showed his true colors by shortening Biltcliffe's suspension.

To be sure, this set of conclusions does not flow ineluctably from the evidence, but it represents a permissible construction of the record. Consequently, the evidence is adequate to support the verdicts against both Roderick and Bogan.

Our journey is not yet ended. The last leg requires us to revisit the lower court's order awarding the plaintiff $83,179.70 in counsel fees and associated expenses against three defendants (Roderick, Bogan, and the City), jointly and severally.

In a section 1983 action a court, "in its discretion, may allow the prevailing party ... a reasonable attorney's fee as part of the costs." 42 U.S.C. § 1988 (1994). Despite its seemingly precatory tone, we have interpreted this language to mean that "a prevailing plaintiff is presumptively entitled to fee-shifting" in a section 1983 case. Casa Marie Hogar Geriatrico, Inc. v. Rivera-Santos, 38 F.3d 615, 618 (1st Cir.1994); accord Foster v. Mydas Assocs., Inc., 943 F.2d 139, 145 (1st Cir.1991) (stating that a prevailing civil rights plaintiff's entitlement to a fee award "comes almost as a matter of course"). For this purpose, a party prevails if she succeeds on a significant issue in the litigation and thereby achieves all or some meaningful part of the benefit that she envisioned when she brought suit. See Hensley v. Eckerhart, 461 U.S. 424, 433, 103 S. Ct. 1933, 1939, 76 L. Ed. 2d 40 (1983); Pearson v. Fair, 980 F.2d 37, 43 (1st Cir.1992). The converse, of course, is equally true: if a plaintiff's claims against a particular defendant come to naught, she is not a prevailing party and is not entitled to reap a harvest under section 1988. See Nunez-Soto v. Alvarado, 956 F.2d 1, 3 (1st Cir.1992). Moreover, if a plaintiff succeeds in the trial court but the judgment she obtains is reversed on appeal, she is no longer entitled to a fee award. See Globe Newspaper Co. v. Beacon Hill Arch. Comm'n, 100 F.3d 175, 195 (1st Cir.1996).

Applying these standards to the case at bar, it is evident that the matter of attorneys' fees must be rethought. Because the plaintiff prevailed below on claims against all three defendants, none of them opposed her application for fees. In their appeals, however, they preserved the issue of whether (and to what extent) the fee award could withstand the reversal on appeal of all or some part of the judgments. This precaution serves the City in good stead; because the judgment against it must be reversed, see supra Part V, the fee award against it must be nullified.

This leaves a nagging question as to the status of the award vis-a-vis Roderick and Bogan. On the one hand, the judgments against those two defendants remain intact, see supra Part VI, and, thus, as to them, the plaintiff remains a prevailing party presumptively entitled to reasonable attorneys' fees. On the other hand, the record before us is opaque as to the proper extent of that entitlement. This opacity is particularly pronounced because we do not know how much (if any) of the work performed by the plaintiff's lawyer in respect to Scott-Harris' unsuccessful claims against the City paved the way for her successful claims against the individual defendants. This is an important datum because a court may allow fees for time spent on unsuccessful claims only if those claims are sufficiently linked to successful claims. See Lipsett v. Blanco, 975 F.2d 934, 940-41 (1st Cir.1992); Aubin v. Fudala, 782 F.2d 287, 290-92 (1st Cir.1986).

We need go no further. From what we have said to this juncture, it is apparent that the matter of fees must be more fully explored--and it is preferable for obvious reasons that the trial court, as opposed to this court, undertake what amounts to an archeological dig into counsel's time sheets and make the necessary factual determinations. We therefore vacate the fee award against the City and remand so that the district court can reconsider the amount of fees and costs that should properly be assessed against the remaining defendants.

The plaintiff also has prevailed on appeal against two of the defendants, and she is entitled to a reasonable counsel fee for the work that yielded this victory. Though we often entertain such fee applications directly, we have sometimes opted to have the district court handle them. See, e.g., Rodi v. Ventetuolo, 941 F.2d 22, 31 (1st Cir.1991); see also 1st Cir.Loc.R. 39.2 (permitting use of this alternative). Because the district court must in any event reopen its inquiry into the overall question of fees, we deem it expedient for the plaintiff to file her application for fees on appeal with that court, and for that court to make the supplementary award. We leave to Judge Saris the procedure to be followed on remand in respect to both reexamination of the original award and initial consideration of the supplementary award for services rendered and expenses (apart from ordinary costs) incurred on appeal.

The plaintiff's cross-appeal (No. 95-2100) is denied and the district court's order permitting the reopening of the appeal period is affirmed. The judgment against the City of Fall River is reversed, and the fee award against it is vacated. The judgments against the remaining defendants are affirmed and the case is remanded to the district court for further proceedings in respect to both the previous fee award against these defendants and the question of fees on appeal. No costs are awarded in Nos. 95-1950 and 95-2100; costs are awarded to the plaintiff in Nos. 95-1951 and 95-1952.