specially concurring:
Judge Smith’s thorough and commendable attempt to navigate through retroactivity jurisprudence with a less than consistent Supreme Court roadmap leaves me in agreement with the result obtained and with some of the reasons for the result. Yet I am sufficiently troubled by other reasoning in the opinion that I have found myself constrained to the options of concurring in the result only or writing a special concurrence. I choose the latter.
I agree with the majority’s assumption that the “Taylor rule” applies to the grand jury context. In my view, the Fourteenth Amendment’s equal protection clause undoubtedly requires a fair cross-section of the community with respect to grand juries. There may be no requirement for states to utilize grand juries, but when they do, this fair cross-section requirement exists. Because I see no legitimate basis for distinguishing between petit and grand juries in this regard, I agree with the majority’s course in this particular case — of assuming that Taylor applies to grand juries.
I cannot, however, join the majority’s position about the nonretroactivity of Griffith. In addition to reversing the law in the Fifth Circuit, the majority’s conclusion that Teague v. Lane bars the retroactive application of the approach to “new rule” retroactivity embraced in Griffith places this Circuit at odds with all of the other circuits that have addressed the issue and with the Supreme Court. The First Circuit in Hill v. Maloney, 927 F.2d 646 (1st Cir.1990), the Eighth in Hamilton v. Jones, 907 F.2d 807 (8th Cir.1990), the Tenth in Liles v. Saffle, 945 F.2d 333 (10th Cir.1991), cert. denied, — U.S. —, 112 S.Ct. 956, 117 L.Ed.2d 123 (1992), and the Eleventh in Pitts v. Cook, 923 F.2d 1568 (11th Cir.1991), all conclude that the Griffith approach to “new rule” retroactivity is to be itself retroactively applied to cases not yet final when the “new rule” in question was announced. And, as the Supreme Court states in Penry:
Penry’s conviction became final on January 13, 1986, when this Court denied his petition for certiorari on direct review of his conviction and sentence. This Court’s decisions in Lockett v. Ohio and Eddings v. Oklahoma were rendered before his con*510viction became final. Under the retroac-tivity principles adopted in Griffith v. Kentucky, Penry is entitled to the benefit of those decisions.
Penry, 492 U.S. 302, 314-315, 109 S.Ct. 2934, 294445, 106 L.Ed.2d 256 (1989) (citations omitted).
The majority takes the position that the Supreme Court does not mean what it says in Penry — because, given that the case rules to which Mr. Penry claimed entitlement were announced before his trial even started, their retroactivity was not really in issue and thus Gñffith was not squarely implicated. Majority Opinion at 507 n. 13 (“Although the Penry Court purported to apply Griffith retroactively, ... it did not appear actually to do so. It granted relief based upon Lockett and Jurek, both of which were decided before Penry’s trial began.”). The majority’s characterization of Penry in this respect is incomplete. It is accurate as far as Lockett and Jurek are concerned. However, the Penry Court gives at least equal billing to Ed-dings — which came down in 1982, after the start of Mr. Penry’s trial but nonetheless before his conviction and sentence became final.
We could of course debate whether the Eddings decision announces a “new rule” or whether it merely reaffirms and refines the rule of Lockett. Essentially, such a debate would be a replay of the one that took place between the majority and dissenting opinions in Eddings itself. But Eddings certainly seems to fit the majority’s understanding of what constitutes a “new rule.” See Majority Opinion at 5699 (“If the conclusion is ‘susceptible to debate among reasonable minds,’ the ... decision is a new rule, even if ‘controlled’ or ‘governed’ by the earlier decision.”) (quoting Butler v. McKellar, 494 U.S. 407, 415, 110 S.Ct. 1212, 1217, 108 L.Ed.2d 347 (1990); and citing Stringer v. Black, — U.S. —, —, 112 S.Ct. 1130, 1140-1141, 117 L.Ed.2d 367 (1992) (Souter, J., dissenting)).
At any rate, the Penry Court relies heavily upon Eddings in order to provide relief to Mr. Penry, and it applies Griffith retroactively in order to do so. If the Penry Court had viewed only Lockett and Jurek as important to its holding, and not Eddings, the Court knew how to say so. Instead, Penry’s plain language clarifies the Court’s intention that Griffith be given retroactive application to habeas petitioners claiming the entitlement to benefit from “new rules” announced before their convictions and sentences became final. I think we must take the Supreme Court at its word, rather than effectively “picking and choosing” the Supreme Court precedents we will and will not faithfully follow.
The majority views this case as presenting a choice between Daniel and Griffith, apparently concluding that the two cases are in conflict. I see no such conflict as these opinions relate to Mr. Wilkerson’s case.
Mr. Wilkerson was in the state’s direct review system, after his jury had been empaneled but before his case had become final, when the Taylor was announced by the Supreme Court. Six days after announcing Taylor, the Court rendered its decision in Daniel — specifically holding that the “Taylor rule” should be applied only to cases in which the juries had not yet been empaneled at the time Taylor was decided. In United States v. Johnson, 457 U.S. 537, 102 S.Ct. 2579, 73 L.Ed.2d 202 (1982), the Supreme Court holds that a defendant’s retroactive entitlement to the benefit of a “new rule” like the one in Taylor is dependent upon whether the “new rule” case at issue has a specific, retroactivity-focused companion among the Supreme Court’s body of precedents. If it does, as it does here (in the form of Daniel), the companion case will govern the retroactivity question. Griffith does not purport to overrule Johnson, and as the majority points out, we are not at liberty to presume that it does so sub silentio.
In my view, the “Griffith” retroactivity approach to “new rule” entitlement claims applies to all eases that were not yet final at the time the “new rule” in question was announced, unless (pursuant to Johnson) the issue of the new rule’s retroactivity is already settled by precedent — that is, by a specific, retroactivity-focused, companion case. It matters not to this analysis whether one raises entitlement to the benefit of a “new rule” by way of direct review or by waf of collateral attack. See e.g., Teague, supra, 489 U.S. at 308-309, 310-311, 109 S.Ct. at *5111074, 1075-76 (1989) (O’Connor, J. (plurality opinion)).
The majority effectively holds that, in any circumstance, Teague bars the retroactive application of the “Griffith” retroactivity approach to eases that are on collateral review. This holding stretches Teague beyond its elastic limits. Teague itself says:
We ... now adopt Justice Harlan’s view of retroactivity for cases on collateral review. [That is,] [u]nless they fall within an exception to the general rule, new constitutional rules of criminal procedure will not be applicable to those cases which have become final before the new rules are announced.
Teague, supra, 489 U.S. at 310-311,109 S.Ct. at 1075 (emphasis added). The most recent cases on point solidify the Supreme Court’s position that habeas courts are to set their “new rule” entitlement sights upon the law as it existed at the time the petitioner’s conviction and sentence became final. See e.g., Lockhart v. Fretwell, — U.S. —, 113 S.Ct. 838, 122 L.Ed.2d 180 (1993). See also Gilmore v. Taylor, — U.S. —, 113 S.Ct. 2112, 124 L.Ed.2d 306 (1993) (holding that subject to two narrow exceptions, a case that is decided after a defendant’s conviction and sentence have become final may not provide the basis for federal habeas relief if that case announces a “new rule”). In one of its most recent reflections upon Teague (the Brecht “harmless error” decision), the Court notes that “new rules” seldom have retroactive application to criminal eases on federal habeas. Brecht v. Abrahamson, — U.S. —, —, 113 S.Ct. 1710, 1720, 123 L.Ed.2d 353 (1993). “Seldom” is not the same as “never.”
The law appears settled. Assuming no specific, retroactivity-focused “companion case” exists to foreclose application, the following represent the “seldom” areas in which “new rules” are to be applied on § 2254 collateral review:
1. cases not yet final when the “new rule” was announced;
2. eases that had become final before the “new rule” was announced, but which concern a “new rule” that places “certain kinds of primary, private individual eon-duct beyond the power of the criminal lawmaking authority to proscribe;”
and
3.cases that had become final before the “new rule” was announced, but which concern a “new rule” requiring the observance of “those procedures that ... are ‘implicit in the concept or ordered liberty.’ ”
See Teague, supra, 489 U.S. at 310-311, 109 S.Ct. at 1075 (“We ... now adopt Justice Harlan’s view of retroactivity for cases on collateral review. [That is,] [u]nless they fall within an exception to the general rule, new constitutional rules of criminal procedure will not be applicable to those cases which have become final before the new rules are announced.”) (emphasis added). See also Teague, supra, 489 U.S. at 307, 109 S.Ct. at 1073 (quoting Mackey v. United States, 401 U.S. 667, 692-693, 91 S.Ct. 1160, 1179-80, 28 L.Ed.2d 404 (1971) (Harlan, J., concurring in judgments in part and dissenting in part); which in turn quotes Palko v. Connecticut, 302 U.S. 319, 325, 58 S.Ct. 149, 152, 82 L.Ed. 288 (1937) (Cardozo, J.)). Of course, “new rule” benefits so applied will still often be subject to Brecht “harmless error” analysis. See Brecht, supra.
Thus, in light of Johnson, supra, I agree with the majority that Daniel prevents Mr. Wilkerson from benefiting from the Taylor rule. I regret that the majority has found it necessary to commit the Fifth Circuit to the lonely and novel position that Teague effectively forecloses the collateral application of the “Griffith ” approach to “new rule” retro-activity questions — even when the “new rule” in issue was announced before the petitioner’s case became final; and even when there is no specific, retroactivity-focused, “companion case” foreclosing such application.