dissenting.
Tennessee v. Garner, 471 U.S. 1, 105 S.Ct. 1694, 85 L.Ed.2d 1 (1985), and our Gamer II opinion which the Supreme Court affirmed, 710 F.2d 240 (6th Cir.1983), limit the authority of the police to shoot to kill. By holding that this fundamental principle of the Gamer opinions applies only prospectively, the majority’s opinion demonstrates a fundamental misunderstanding of the principle of retroactivity and creates an unnecessary conflict in the circuits on an important constitutional issue. The majority basically looks only to *1138Chevron Oil Co. v. Huson, 404 U.S. 97, 106-07, 92 S.Ct. 349, 355, 30 L.Ed.2d 296 (1971), and then gives it an erroneous interpretation. This view reveals a lack of understanding of the general principle that judicial decisions have retroactive effect. See Goodman v. Lukens Steel Co., — U.S. -, 107 S.Ct. 2617, 2621, 96 L.Ed.2d 572 (1987). The majority ignores the “usual rule” that federal cases are decided “in accordance with the law existing at the time of decision.” Id.; see also Saint Francis College v. Al-Khazraji, — U.S. -, 107 S.Ct. 2022, 2025, 95 L.Ed.2d 582 (1987); Gulf Offshore Co. v. Mobil Oil Corp., 453 U.S. 473, 486 n. 16, 101 S.Ct. 2870, 2879 n. 16, 69 L.Ed.2d 784 (1981); Thorpe v. Housing Authority, 393 U.S. 268, 281, 89 S.Ct. 518, 526, 21 L.Ed.2d 474 (1969); United States v. The Schooner Peggy, 1 Cranch 103, 109, 2 L.Ed. 49 (1801).
For hundreds of years the common law applied the current rule of law to pending cases, and thereby established a strong presumption that all judicial decisions apply retroactively.1 This presumption grew out of the view that judges were not creators but discoverers of the law. Linkletter v. Walker, 381 U.S. 618, 622, 85 S.Ct. 1731, 1733, 14 L.Ed.2d 601 (1965). As Blackstone explained, judges were “not delegated to pronounce a new law, but to maintain and expound the old one.” 1 W. Blackstone, Commentaries 69 (1765). A new ruling did not mean that a contrary old rule was “bad law, but that it was not law.” Commentaries at 70 (emphasis in original). Chief Justice Marshall made the same point:
[I]f, subsequent to the judgment [in the trial court] and before the decision of the appellate court, a law intervenes and positively changes the rule which governs, ... the court must decide according to existing laws, and if it be necessary to set aside a judgment ... which cannot be affirmed, but in violation of law, the judgment must be set aside.
The Schooner Peggy, 1 Cranch at 110. To the same effect, Justice Holmes:
I know of no authority in this court to say that in general state decisions shall make law only for the future. Judicial decisions have had retrospective operation for near a thousand years.
Kuhn v. Fairmont Coal Co., 215 U.S. 349, 372, 30 S.Ct. 140, 148, 54 L.Ed. 228 (1910) (dissenting). And Chief Justice Rehnquist:
The principle that statutes operate only prospectively, while judicial decisions operate retrospectively, is familiar to every law student.
United States v. Security Industrial Bank, 459 U.S. 70, 79, 103 S.Ct. 407, 413, 74 L.Ed.2d 629 (1982).
The majority opinion presents a distorted picture both of the decisional evolution of the Supreme Court ruling in Gamer and of the legal standard that must be met to justify the exceptional step of applying a judicially determined rule of law nonretro-actively. Specifically, the majority opinion:
(1) mischaracterizes the relationship between the Sixth Circuit and Supreme Court decisions in Gamer and the earlier “fleeing felon” precedents;
(2) fails to recognize that under Owen v. City of Independence, 445 U.S. 622, 100 S.Ct. 1398, 63 L.Ed.2d 673 (1980), the narrow exceptions permitted by Chevron must be narrowed further in a case involving a municipal defendant and a constitutional issue, thus providing additional reason to apply Garner retroactively;
(3) misstates the principal purposes that underlie the litigation of constitutional torts under § 1983 and exaggerates the “hardship” that might result from requiring municipal governments to be sensitive to developing constitutional trends; and
(4) creates a conflict with the rule of retroactivity adopted, either expressly or implicitly, and applied by seven other circuits. It also ignores the inequity that results when similarly situated victims of constitutional deprivations are treated differently due to inconsistent and unprincipled application of nonretroactivity.
*1139I.
The majority’s first task is to establish that the defendant in this case — now only the City of Chattanooga — should have relied on the validity of the Tennessee statute permitting the use of deadly force to apprehend unarmed, nonviolent felony suspects. This reliance factor, as described by Chevron, requires that:
[T]he decision to be applied nonretroac-tively must establish a new principle of law, either by overruling clear past precedent on which litigants have relied, see, e.g., Hanover Shoe v. United Shoe Machinery Corp., [392 U.S. 481] at 496, [88 S.Ct. 2224, 2233, 20 L.Ed.2d 1231] or by deciding an issue of first impression whose resolution was not clearly foreshadowed, see, e.g., Allen v. State Board of Elections, [393 U.S. 544] at 572, [89 S.Ct. 817, 835, 22 L.Ed.2d 1].
404 U.S. at 106, 92 S.Ct. at 355. The majority’s explanation of how this language applies to the Gamer principle suffers from two critically important defects. ******
First, Gamer did not overrule a clear past precedent, either in the Supreme Court or the Court of Appeals, on which the City of Chattanooga reasonably could have relied. Because officer Kyle shot Adrian Carter on December 21, 1982, we examine prior case law established as of that date in order to determine if the City relied on a clear past precedent to justify its use of deadly force. On December 21, 1982, there was no Supreme Court opinion on the subject, and the following six cases relating to Tennessee’s fleeing felon statute had been decided by federal courts: Cunningham v. Ellington, 323 F.Supp. 1072 (W.D.Tenn.1971); Beech v. Melancon, 465 F.2d 425 (6th Cir.1972), cert. denied, 409 U.S. 1114, 93 S.Ct. 927, 34 L.Ed.2d 696 (1973); Qualls v. Parrish, 534 F.2d 690 (6th Cir.1976); Wiley v. Memphis Police Dept., 548 F.2d 1247 (6th Cir.), cert. denied, 434 U.S. 822, 98 S.Ct. 65, 54 L.Ed.2d 78 (1977); Garner I, 600 F.2d 52 (6th Cir.1979); Haislah v. Walton, 676 F.2d 208 (6th Cir.1982).
An examination of these cases reveals that no past precedent had established that Tennessee’s fleeing felon statute was constitutional on the basis of the Fourth Amendment. And two of these cases clearly indicated that the constitutionality of the statute was an open question.
Cunningham, Beech, Qualls and Wiley each discussed Tennessee’s fleeing felon statute, but none of these cases focused on the issue of Gamer: whether such police conduct may violate the Fourth Amendment. As our 1983 panel opinion in Garner II said, the “question under the Fourth Amendment is one of first impression in this Court.” 710 F.2d at 243. See Cunningham, 323 F.Supp. at 1075-76 (statute not unconstitutional on its face under Eighth and Fourteenth Amendments); Beech, 465 F.2d at 426 (relying on Cunningham, statute not unconstitutional; Fourth Amendment not discussed); Qualls, 534 F.2d at 694 (police officer justified to believe deadly force necessary; no constitutional issue discussed). Wiley is the only case in which the Fourth Amendment was even mentioned, and then only as one of six separate constitutional theories involved on a jurisdictional basis. 548 F.2d at 1248. The Wiley opinion, however, contained no discussion of the Fourth Amendment claim and relied only on Cunningham and Beech, both due process holdings. Because it was decided before Monell v. Dept. of Soc. Servs., 436 U.S. 658, 98 S.Ct. 2018, 56 L.Ed.2d 611 (1978), Wiley did not address municipal liability and held only that the individual officers were entitled to good faith immunity because they “had a right to assume that the statute was constitutional.” Id. at 1251 (emphasis added). Wiley did not hold that the Tennessee statute passed Fourth Amendment muster.
The majority attempts to justify its assertion that Gamer I contained no “warning signal” by a long and detailed recital of the failure of Judge Wellford, then sitting as a district judge, to discern that signal upon remand in Gamer I. Two other district judges, however, did not find the message of Gamer I so difficult to understand. See, e.g., Taylor v. Collins, 574 F.Supp. 1554, 1558 (E.D.Mich.1983) (Garner I foretold Sixth Circuit’s decision in *1140Garner II); Jacobs v. City of Wichita, 531 F.Supp. 129, 132 (D.Kan.1982) (after Garner I, the Sixth Circuit no longer “reads its prior opinions as based on ... the constitutionality of the common law rule”). In Taylor, in which the District Court concluded that “retroactive application of Garner II is appropriate,” 574 F.Supp. at 1559, Judge Newblatt explained that a law review comment cited in Gamer I “reveals that the historical foundation of American state fleeing-felon statutes is a foundation built on loose sand." Id. at 1558. Judge Newblatt concluded that after Garner I, “the Garner II Fourth Amendment ruling ... was no shock.” Id. It seems to me that the plain meaning of the Court’s language in Gamer 1 clearly signaled in 1979 to various cities, including Chattanooga, that Tennessee’s statute as applied to unarmed, nonviolent felony suspects, might violate the Fourth Amendment. Moreover, if there were any doubt about the meaning of Garner I, the Sixth Circuit repeated the warning again in an opinion published eight months before Carter was shot in Chattanooga. See Haislah v. Walton, 676 F.2d 208 (April 21, 1982).
In Haislah, another case involving a fleeing suspect, the Sixth Circuit reversed a jury decision for the city and a police officer in part because the district court’s instructions misleadingly “implied that the absence of [police officer] Walton’s liability meant that ... no constitutional violation has been committed.” 676 F.2d at 214 n. 3. The court explained:
When the asserted justification is the apprehension of a fleeing felon, however, the exoneration of the police officer does not necessarily settle the constitutionality of his conduct or the liability of his governmental employer. See Garner [I], 600 F.2d 52, 54-55 (6th Cir.1979).
Id.
Could it have been plainer that the Sixth Circuit viewed the constitutionality of the fleeing felon statutes as in some doubt?
Thus, there was no “clear past precedent” in Cunningham, Beech, Qualls, and Wiley, none of which involved the Fourth Amendment, and each of which centered on the issue of good faith immunity for the individual officer. And the majority’s disregard of the clear signal in Garner I, a signal repeated again in Haislah, reveals how result-oriented the majority’s mission here is.
For years many have considered the type of police killings condoned here morally wrong and constitutionally suspect. The majority is wrong that Gamer created a “novel” rule. Almost twenty years ago former Chief Justice Warren Burger said in Bivens v. Six Unknown Named Agents, 403 U.S. 388, 419, 91 S.Ct. 1999, 2016, 29 L.Ed.2d 619 (1971), “that a ‘shoot’ order might conceivably be tolerable to prevent the escape of a convicted killer but surely not for car thieves, pickpockets or a shoplifter.” And as we noted in Gamer II, 710 F.2d at 246, the Sixth Circuit more than 100 years ago said that an officer cannot legally kill an unarmed, nonviolent, fleeing suspect:
Suppose, for example, a person were arrested for petit larceny, which is a felony at the common law, might an officer under any circumstances be justified in killing him? I think not. The punishment is altogether too disproportionate to the magnitude of the offense.
United States v. Clark, 31 F. 710, 713 (1887).
Moreover, the majority ignores the Supreme Court’s historical summary in Tennessee v. Garner that “the long-term movement has been away from the rule that deadly force may be used against any fleeing felon.” 471 U.S. at 18, 105 S.Ct. at 1705. The Court reviewed studies reporting that “only 7.5% of departmental and municipal policies explicitly permit the use of deadly force against any felon; 86.8% explicitly do not.” Id. at 19, 105 S.Ct. at 1705. The Court noted that the Model Penal Code rejected the old common law rule. The Court concluded that, in view of the more enlightened policies adopted by the police themselves, “the older and fading common-law view is a dubious indicium” of the continuing constitutionality of the Tennessee statute. Id.
*1141The majority gives scant attention to two recent Eleventh Circuit cases in which separate unanimous panels explicitly ruled that under Chevron, Gamer should apply retroactively. Acoff v. Abston, 762 F.2d 1543, 1548-49 (11th Cir.1985); Pruitt v. City of Montgomery, 771 F.2d 1475, 1478-83 (11th Cir.1985); see Maj. opin. at 1133 n. 11, 1135 n. 12. In Acoff, Judge Johnson explained:
We hold that the Gamer decision was not an entirely new and unanticipated principle of law that would justify nonre-troactivity. The decision, while not a simple application of past precedent, did follow the “balancing” methodology of many recent Supreme Court opinions.... There was no prior precedent overruled by Gamer. Nor does the decision apply to any longstanding and widespread practice which the Supreme Court had sanctioned in prior cases or ignored while lower courts had approved of the practice with near unanimity. Indeed, several courts had expressed the view that certain uses of deadly force such as the one in question here violate the Constitution.
762 F.2d at 1549 (citations omitted).
In two recent Supreme Court cases, Goodman v. Lukens Steel Co., — U.S. -, 107 S.Ct. 2617, 96 L.Ed.2d 572 (1987), and Saint Francis College v. Al-Khazraji, — U.S. -, 107 S.Ct. 2022, 95 L.Ed.2d 582 (1987), the Court provided a vivid illustration of the importance of “clear past precedent” in the retroactivity determination under Chevron. In Saint Francis, the Court upheld the nonretroactive application of Wilson v. Garcia, 471 U.S. 261, 105 S.Ct. 1938, 85 L.Ed.2d 254 (1985), because when plaintiff filed his suit, it was clearly established in the Third Circuit that a § 1981 plaintiff must comply with a six year statute of limitations. 107 S.Ct. at 2025-26. In Goodman, however, the Supreme Court ruled that because there was no clear past precedent relating to the statute of limitations on which a plaintiff could rely, Wilson v. Garcia should apply retroactively. 107 S.Ct. at 2621-22.
The two cases involved the application of a statute of limitations for the same cause of action, 42 U.S.C. § 1981, in the same (Third) Circuit. The difference between the two cases was that the plaintiff in St. Francis filed his lawsuit after two Third Circuit cases had expressly held that the six year limitation applied. The Goodman plaintiffs suit was filed before those two cases, at a time when there was no “established precedent in the Third Circuit,” the law was “unsettled,” and thus a decision “should be given the customary retroactive effect.” Goodman, 107 S.Ct. at 2622. Thus, the Wilson v. Garcia Supreme Court holding was applied retroactively in the same circuit to a case filed in 1973 but not to a case filed in 1980. The difference obviously was not time, but precedent and the strength of the reliance interest at stake.
In Garner, we are dealing not with a question of which state statute of limitations should apply, a narrow, specific question of federal common law, but rather with a broad constitutional question under the Fourth Amendment which had not been conclusively decided in the Supreme Court or elsewhere. See Bradley v. Richmond School Board, 416 U.S. 696, 719, 94 S.Ct. 2006, 2020, 40 L.Ed.2d 476 (1974).
Moreover, the reliance question here is how the Supreme Court’s decision in Garner, not this Court’s decision, should be applied. It is a strange constitutional principle of retroactivity that prohibits police shootings in Alabama but permits them in Tennessee. Nothing in this circuit’s line of cases creates the kind of strong reliance interest that might justify such an anomalous result.
♦ * * * * *
The second major flaw in the majority’s analysis of Chevron’s reliance factor is its failure to understand the narrowing effect that Owen v. City of Independence, 445 U.S. 622, 100 S.Ct. 1398, 63 L.Ed.2d 673 (1980), must have on nonretroactivity in cases of municipal liability. In Owen, the City had discharged the plaintiff two months before the Supreme Court had issued its opinions in two cases that served *1142to establish the Owen plaintiffs right to a hearing. The district court had held that the City could not be “reasonably ... charged with constructive notice of such rights.” Id. at 631 n. 10, 100 S.Ct. at 1405 n. 10. But the Supreme Court flatly rejected the City’s argument that it
could not have been aware of [petitioner’s] right to a name-clearing hearing in connection with the discharge [and] should not be charged with predicting the future course of constitutional law.
Id. at 634, 100 S.Ct. at 1407.2 The Court held that “municipalities have no immunity from damages liability flowing from their constitutional violations.” Id. at 657, 100 S.Ct. at 1418. And, the Court explained:
[E]ven where some constitutional development could not have been foreseen by municipal officials, it is fairer to allocate any resulting financial loss to the inevitable costs of government borne by all the taxpayers, than to allow its impact to be felt solely by those whose rights, albeit newly recognized, have been violated.
Id. at 655, 100 S.Ct. at 1417 (emphasis added).
Finally, the Court announced that it wanted to “create an incentive for officials who may harbor doubts about the lawfulness of their intended actions to err on the side of protecting citizens’ constitutional rights.” Id. at 651-52, 100 S.Ct. at 1416 (emphasis added).
The Chevron analysis must be carried out in the clear light of Owen. Although Owen is about qualified immunity and Chevron is about nonretroactivity, both are squarely focused on the issue of reliance that is common and crucial to both doctrines. Surely when we analyze whether to apply a new basis for constitutional liability nonretroactively, we should not exempt a city from liability under Chevron if that same city should be liable under the express holding of Owen.
Neither do I think that the phrase “an issue of first impression whose resolution was not clearly foreshadowed,” Chevron, 404 U.S. at 106, 92 S.Ct. at 355, should be applicable to this type of § 1983 case, involving a constitutional issue and a city defendant. Even apart from the narrowing effect of Owen, the mere fact that a case is one of first impression whose result has not clearly been foreshadowed ordinarily does not, by itself, suffice to trigger nonretroactivity. The few Supreme Court cases of this “first impression” type all contained other compelling reasons to bar retroactive application, see Northern Pipeline Construction Co. v. Marathon Pipe Line Co., 458 U.S. 50, 102 S.Ct. 2858, 73 L.Ed.2d 598 (1982) (new rule applied prospectively to avoid substantial injustice and hardship against thousands of litigants who relied on jurisdiction of bankruptcy courts); Cipriano v. City of Houma, 395 U.S. 701, 706, 89 S.Ct. 1897, 1900, 23 L.Ed. 2d 647 (1968) (substantial inequity and hardships would result if elections voided; decision applies prospectively); Allen v. State Board of Elections, 393 U.S. 544, 571-72, 89 S.Ct. 817, 834-35, 22 L.Ed.2d 1 (1968) (elections not voided; new rule applies prospectively). In each of these cases, the Supreme Court applied new judicial decisions nonretroactively to avoid substantial hardship, inequitable results, or massive disruption. Here, retroactive application of Gamer hardly approaches the impact of disrupting thousands of bankruptcy decisions or voiding elections already held.
Therefore, the fact that the Gamer Fourth Amendment holding was one of “first impression” is of no moment to the Chevron reliance analysis. The only inquiry necessary is whether the city relied on a clear past precedent. If the reliance was not based on a clear past precedent, the city’s reliance was not justified and therefore the new rule must apply retroactively.
In this case, there existed no precedent holding that the fleeing felon statute complied with the Fourth Amendment, and two *1143Sixth Circuit cases had declared the law in this area unsettled. Likewise, the law of the nation as a whole was not settled. The Supreme Court had never spoken on the question. Chattanooga certainly should have harbored some doubt about the constitutionality of its fleeing felon policy, and could not have reasonably relied on existing law to justify its authorization of deadly force.
II.
The majority’s cryptic discussion of Chevron’s second factor is also troublesome. Under Chevron the court must:
weigh the merits and demerits in each case by looking to the prior history of the rule in question, its purpose and effect, and whether retrospective operation will further or retard its application.
404 U.S. at 106-07, 92 S.Ct. at 355. The majority not only distorts Gamer’s effect on deterring cities from unconstitutional policies, it also virtually ignores Garner’s remedial purpose of providing compensation.
In the majority’s view, the primary purpose of Gamer — a purpose the majority obviously considers of dubious value — is to deter cities from certain policies sanctioning deadly force in violation of the Fourth Amendment. Because it is too late to deter past policies or conduct, the majority concludes we should not apply Gamer retroactively. The opinion states:
Modifying one’s behavior to comply with a future change of a rule, particularly an unforeseeable change, is difficult if not impossible. Retroactive application of Gamer would, therefore, have little, if any, effect of furthering the deterrent goal of Garner.
Maj. opin. at 1130.
By claiming that the purpose of Gamer is to deter the city from violating only those legal standards that are already crystal clear, the majority has erected a “stand pat” approach that is flatly contrary to the stated purpose of the Civil Rights Act and once again flatly contrary to the holding of the Supreme Court in Owen v. City of Independence. It is true that one cannot deter the performance of an illegal act already committed. But the logical outcome of the majority’s approach would be the adoption of a uniform rule that virtually all judicial decisions apply prospectively only. Under this approach, the Supreme Court should not have applied the new Gamer rule to Gamer itself, and except in “easy” cases no court would apply its instant holding to the instant facts. The majority’s error is demonstrably fundamental, and as Justice Holmes pointed out in the Kuhn case, 215 U.S. at 372, 30 S.Ct. at 148, contrary to “near a thousand years” of Anglo-American jurisprudence.
The majority’s view that deterrence must be limited to only foreseeable changes in the law has a superficial appeal, reminding one of the concepts of duty and proximate cause in tort law. But the Supreme Court rejected this formulation, as discussed at length above, in Owen. Cities are to be held liable for, and thus be deterred from, violations even when the constitutional rights in question are “newly recognized” and the result of “some constitutional development [which] could not have been foreseen by municipal officials.” 445 U.S. at 655, 100 S.Ct. at 1417 (emphasis added).
The Supreme Court in Owen also recognized that § 1983 actions have two main purposes, not only to deter against future constitutional deprivations, but also “to provide compensation to the victims of past abuses,” 445 U.S. at 651, 100 S.Ct. at 1416 (citing Carey v. Piphus, 435 U.S. 247, 256-57, 98 S.Ct. 1042, 1048-49, 55 L.Ed.2d 252 (1978); Robertson v. Wegmann, 436 U.S. 584, 590-91, 98 S.Ct. 1991, 1995, 56 L.Ed.2d 554 (1978)). The majority virtually ignores Gamer’s effect on a suspect’s constitutional rights and only grudgingly acknowledges that victims of such violations are entitled to receive compensation. Retroactive application of Garner would make Chattanooga understand the seriousness of violating a suspect’s Fourth Amendment rights through deadly police force.
The majority makes the remarkable assertion that compensation of victims of constitutional deprivations is “merely a consequence” of a § 1983 judgment applying the *1144Garner principle. Maj. opin. at 1130. It omits the actual language from Owen describing the essential purposes of § 1983 actions:
A damages remedy against the offending party is a vital component of any scheme for vindicating cherished constitutional guarantees, and the importance of assuring its efficacy is only accentuated when the wrongdoer is the institution that has been established to protect the very rights it has transgressed.
445 U.S. at 651, 100 S.Ct. at 1415. The Court also explained:
The knowledge that a municipality will be liable for all of its injurious conduct, whether committed in good faith or not, should create an incentive for officials who may harbor doubts about the lawfulness of their intended actions to err on the side of protecting citizens’ constitutional rights. Furthermore, the threat that damages might be levied against the city may encourage those in a policymaking position to institute internal rules and programs designed to minimize the likelihood of unintentional infringements on constitutional rights. Such procedures are particularly beneficial in preventing those “systemic” injuries that result not so much from the conduct of any single individual, but from the interactive behavior of several government officials, each of whom may be acting in good faith.
Id. at 651-52, 100 S.Ct. at 1416 (footnotes omitted) (emphasis added); see also Mish-kin, Foreword: The High Court, the Great Writ, and the Due Process of Time and Law, 79 Harv.L.Rev. 56, 72 (1965).
The majority also does not appreciate that recovery of damages is essentially the only means of enforcing the Gamer rule. In most other cases in which the constitutional deprivation is a matter of criminal procedure, enforcement is primarily guaranteed by operation of the exclusionary rule, see Payton v. New York, 445 U.S. 573, 100 S.Ct. 1371, 63 L.Ed.2d 639 (1979); Mapp v. Ohio, 367 U.S. 643, 81 S.Ct. 1684, 6 L.Ed.2d 1081 (1961); Johnson v. United States, 333 U.S. 10, 68 S.Ct. 367, 92 L.Ed. 436 (1948). In Elkins v. United States, 364 U.S. 206, 217, 80 S.Ct. 1437, 1444, 4 L.Ed.2d 1669 (1959), the Supreme Court explained that the exclusionary rule
is calculated to prevent, not to repair. Its purpose is to deter — to compel respect for the constitutional guaranty in the only effectively available way — by removing the incentive to disregard it.
But in situations like Gamer, usually no tangible evidence or statements are retrieved as a result of the constitutional violation. Thus, the usual threat of the exclusionary rule does not,help guarantee compliance with the Supreme Court’s rule that deadly force violates an unarmed, nonviolent felony suspect’s Fourth Amendment rights. Without an enforcement mechanism, the Court has no effective means of requiring compliance with constitutional decisions. In Gamer cases, then, the remaining enforcement mechanism is the threat of an action for damages under § 1983 against cities with policies sanctioning the use of deadly force against unarmed, nonviolent felony suspects.
III.
Finally, Chevron requires a court to weigh:
the inequity imposed by retroactive application, for “[wjhere a decision of this Court could produce substantial inequitable results if applied retroactively, there is ample basis in our cases for avoiding the ‘injustice or hardship’ by a holding of nonretroactivity.” Cipriano v. City of Houma, [395 U.S.] at 706.
404 U.S. at 107, 92 S.Ct. at 355 (emphasis added). As already discussed above, applying the Gamer rule retroactively does not at all implicate the types of hardship or disruption that sometimes have sufficed to justify nonretroactivity.
In considering equity and consistency in the application of the law, we note that the Supreme Court applied the Gamer rule in the Gamer case. The shooting of Edward Garner occurred October 3, 1974; the shooting of Adrian Carter occurred December 21, 1982. More difficult to conceive than anything else said by my colleagues in *1145the majority is their statement that “weighing and balancing of the equities in this case,” Maj. opin. at 1130, requires denying recovery to the mother of this dead boy who, though unarmed and on the run was shot in the back in violation of the Fourth Amendment of the Constitution of the United States. There is nothing in any of the opinions in Gamer quite so “novel” as that.
It is striking that no less than seven other circuits have applied Gamer retroactively. Fernandez v. Leonard, 784 F.2d 1209, 1217 (1st Cir.1986) (under Garner, police officer violated suspect’s Fourth Amendment rights on December 27, 1976); Kidd v. O’Neil, 774 F.2d 1252, 1255-56 (4th Cir.1985) (under Gamer, summary judgment reversed where police used excessive force against suspect on April 15, 1983); Griffin v. Hilke, 804 F.2d 1052, 1055 (8th Cir.1986), cert. denied, — U.S. -, 107 S.Ct. 3185, 96 L.Ed.2d 673 (1987) (Gamer applies to § 1983 action where suspect shot on February 18, 1978); Ryder v. City of Topeka, 814 F.2d 1412, 1417-18 (10th Cir.1987) (Garner applies to suit where fleeing suspect shot by police on March 25, 1979); Pmitt v. City of Montgomery, 771 F.2d 1475, 1478-79 (11th Cir.1985) {Gamer applies to § 1983 action where plaintiff shot by police on September 1, 1982); see also Jamieson v. Shaw, 772 F.2d 1205, 1209-10 (5th Cir.1985) (District Court should have permitted plaintiff suspect to amend complaint with Fourth Amendment claim under Gamer for injury on July 25, 1982); Bell v. City of Milwaukee, 746 F.2d 1205, 1278 n. 87 (7th Cir.1984) (Garner II applies to § 1983 action where fleeing suspect shot by police on February 2, 1958). In these cases, courts did not expressly analyze whether Gamer should have retroactive effect. They assumed it. Despite the majority’s lengthy attempt to distinguish the facts of these cases from Gamer and Carter, it does not — and cannot — deny that these courts did indeed apply the Gamer holding to these earlier incidents.
The Supreme Court’s decision on the Tennessee statute at issue in Gamer had obvious implications for the 18 other states which also retained codifications of the old common law fleeing felon rule. On issues of “great national concern,” the presumption of retroactivity must be applied with more force than usual. Bradley v. Richmond School Board, 416 U.S. 696, 719, 94 S.Ct. 2006, 2020, 40 L.Ed.2d 476 (1974) (applying liability for attorney fees in desegregation case retroactively against municipal school board). Across the country many other past victims of excessive police force are being recompensed, and the city defendants are not even arguing the issue of Garner’s retroactivity. As explained supra, pp. 1141, the Eleventh Circuit in Acoff v. Abston, 762 F.2d 1543, 1548-49 (11th Cir.1985), fully analyzed whether Gamer should apply retroactively and concluded that it should.3
Basically, what has happened in this en banc case is that the majority — in its hostility to the liberalization of the law by the Supreme Court and in its zeal to show that the panel decisions of this Court in Gamer I and Gamer II were wrong — has forgotten that the question before us concerns the retroactivity of the Supreme Court’s opinion in Gamer. No other court has agreed with the majority, and every sensi*1146ble argument leads to the opposite conclusion.
In conclusion, the holding in Gamer should have been applied retroactively as a matter of law to the facts of this case. The District Judge did so when he ruled as a matter of law that the Chattanooga policy was unconstitutional and when he charged the jury that Kyle was acting in conformity with municipal policy. However, his instructions to the jurors allowed them to engage in a redetermination of the “constitutionality” or “unconstitutionality” of the City’s policy insofar as it treated all burglars as potentially dangerous. App. 140-51, 191. The jurors’ verdict for the City probably resulted from a decision that the City’s policy was constitutional, but it may have resulted from a conclusion that Kyle had the requisite probable cause to shoot. If the former, the verdict was contrary to law. If the latter, it was a verdict that no reasonable jury could have reached, given the testimony of Officer Kyle that he shot Carter solely because he was a fleeing burglar. App. 77-78, 103-04. The District Court should have granted plaintiff’s motion for judgment notwithstanding the verdict.
. Like Chattanooga, the City of Independence in Owen relied on prior case law which did not establish a clear precedent that the city’s actions violated a constitutional right. See 445 U.S. at 633 n. 13, 634, 100 S.Ct. at 1406 n. 13, 1407.
. In Griffith v. Kentucky, 479 U.S. 314, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987), the Supreme Court ruled that new criminal procedural rules automatically apply retroactively to pending cases on direct appeal. Griffith may provide some support for the view that Gamer should automatically apply retroactively because the duty to respect suspects' Gamer rights arises from a Fourth Amendment principle which finds its most common application in criminal procedure. Consistency may require that Griffith apply to civil cases, such as this one, arising out of criminal episodes. The extent of the applicability of Griffith in the civil context is an open question, and there are difficult counterarguments. For example, in the criminal area, the Griffith rule of automatic retroactivity applies only to cases on direct review. Allen v. Hardy, 478 U.S. 255, 106 S.Ct. 2878, 92 L.Ed.2d 199 (1986), continues to hold that a new rule of criminal procedure is applied retroactively on collateral review in habeas corpus cases only if it has “a fundamental impact on the integrity of factfinding." Id. at 2878. Thus the question of Griffith's applicability remains unresolved; in light of our discussion above, we need not address it further.