A state prisoner named Richardson appeals from the denial of his application for habeas corpus, arguing that the prosecutor used a peremptory challenge to strike a juror by reason of the juror's race in violation of the constitutional principle declared in Batson v. Kentucky, 476 U.S. 79, 106 S. Ct. 1712, 90 L. Ed. 2d 69 (1986). Richardson can use Batson to attack his conviction only if the conviction was not yet "final" when Batson was decided. Allen v. Hardy, 478 U.S. 255, 106 S. Ct. 2878, 92 L. Ed. 2d 199 (1986). A criminal judgment is final for this purpose when the defendant has exhausted his state appellate remedies and either the U.S. Supreme Court has denied his petition for certiorari or the time for filing the petition has expired. Id. at 258 n. 1, 106 S. Ct. at 2879 n. 1; Teague v. Lane, 820 F.2d 832, 836 (7th Cir.1987) (en banc), aff'd, 489 U.S. 288, 109 S. Ct. 1060, 103 L. Ed. 2d 334 (1989); Byrd v. Armontrout, 880 F.2d 1, 6-7 (8th Cir.1989).

Richardson was convicted of murder and robbery in an Illinois state court in 1984 and sentenced to sixty years in prison. He appealed, raising the Batson issue albeit in pre-Batson form--Batson not yet having been decided--but adequately to preserve the issue. Trevino v. Texas, --- U.S. ----, ----, 112 S. Ct. 1547, 1550, 118 L. Ed. 2d 193 (1992). On December 30, 1985, the Illinois Appellate Court rendered a decision in which it "affirmed" the "judgment" of the trial court but "remanded for resentencing." People v. Richardson, 139 Ill.App.3d 598, 603, 93 Ill. Dec. 891, 894, 487 N.E.2d 716, 719 (1985). The reason for the remand was as follows. After Richardson had been convicted and sentenced, two other men had been convicted as participants in the same offense and the evidence at the trial of one of them, Stewart, indicated that Richardson had been an accessory rather than one of the principals, as the judge who sentenced Richardson had thought. The appellate court pointed out that "the sentencing judge could not possibly have known of the evidence which would later be adduced at Stewart's trial when he sentenced defendant, so it cannot be said that he erred in sentencing defendant." Id. Nevertheless, to avoid injustice as well as to economize on judicial resources the court remanded the case so that the judge could resentence Richardson in light of the evidence that had surfaced in the subsequent trial. Richardson did not seek leave to appeal to the Supreme Court of Illinois or file a petition for certiorari in the U.S. Supreme Court.

On May 22, 1986, the judge resentenced Richardson to thirty years, thus cutting the original sentence in half. The Illinois Appellate Court affirmed the new sentence on November 23, 1987, and the Supreme Court of Illinois denied leave to appeal. Batson had been decided by the U.S. Supreme Court on April 30, 1986. That was after the Illinois Appellate Court's decision affirming the judgment of conviction had been rendered and after the time for Richardson to seek leave to appeal to the Supreme Court of Illinois or to file a petition for certiorari in the U.S. Supreme Court had expired, but before Richardson was resentenced and almost two years before the judgment imposing the new sentence became final.

After the state supreme court denied leave to appeal in the second round, Richardson mounted a collateral attack in the state court system on his conviction. He was denied relief, on the ground that his conviction had become final before Batson was decided. He then brought this federal habeas corpus action.

A judgment is not final if the appellate court has remanded the case to the lower court for further proceedings, unless the remand is for a purely "ministerial" purpose, involving no discretion, such as recomputing prejudgment interest according to a set formula. United States v. F. & M. Schaefer Brewing Co., 356 U.S. 227, 233-34, 78 S. Ct. 674, 678, 2 L. Ed. 2d 721 (1958); Herzog Contracting Corp. v. McGowen Corp., 976 F.2d 1062, 1064 (7th Cir.1992); Parks v. Pavkovic, 753 F.2d 1397, 1401-02 (7th Cir.1985); In re Riggsby, 745 F.2d 1153, 1156 (7th Cir.1984); Freeman United Coal Mining Co. v. Director, 721 F.2d 629, 631 (7th Cir.1983); Morgan v. United States, 968 F.2d 200, 204-05 (2d Cir.1992); In re Gould & Eberhardt Gear Machinery Corp., 852 F.2d 26, 29 (1st Cir.1988). It is not final, therefore, if it merely determines liability, leaving the determination of relief to subsequent proceedings. Liberty Mutual Ins. Co. v. Wetzel, 424 U.S. 737, 744, 96 S. Ct. 1202, 1206, 47 L. Ed. 2d 435 (1976); In re Fox, 762 F.2d 54, 55 (7th Cir.1985); In re Goldblatt Bros., 758 F.2d 1248, 1250 (7th Cir.1985). So, in a criminal case, a judgment of conviction is not final, because the "relief" sought by the prosecution is not the conviction as such but the sentence. Berman v. United States, 302 U.S. 211, 212, 58 S. Ct. 164, 165, 82 L. Ed. 204 (1937); Flanagan v. United States, 465 U.S. 259, 263, 104 S. Ct. 1051, 1053, 79 L. Ed. 2d 288 (1984); United States v. Patel, 835 F.2d 708, 709 (7th Cir.1987).

This analysis might seem to establish beyond the possibility of doubt that the Illinois Appellate Court's decision of December 30, 1985, while purporting to affirm the trial court's judgment, was not final, because the case was remanded for resentencing in the sentencing judge's discretion rather than pursuant to some mechanically applicable rule. But there are two reasons to doubt this conclusion. The first is that the decision did not purport to vacate the sentence. Compare United States v. Moree, 928 F.2d 654, 656 (5th Cir.1991). It remanded the case so that the sentencing judge "can [if he wants] resentence defendant to a term of imprisonment appropriate in light of the subsequent events." People v. Richardson, supra, 139 Ill.App.3d at 603, 93 Ill. Dec. at 894, 487 N.E.2d at 719 (emphasis added). The opinion ends with the words "Affirmed and Remanded"--as if the contemplated resentencing were a collateral proceeding, such as an action for habeas corpus, which normally follows the affirmance of the defendant's conviction. The court was worried that if the sentence "is affirmed [without a remand], it may be that defendant will be unable to obtain sentencing relief to which he may be entitled." Id. Perhaps some hitch in the state's system of postconviction remedies would have prevented Richardson from obtaining a correction of his sentence in light of the newly discovered evidence presented at Stewart's trial, so that the appellate court was in effect directing the sentencing judge to reopen the case, just as if, in a federal civil trial, Richardson had moved under Fed.R.Civ.P. 60(b) for a new trial on the basis of newly discovered evidence. The filing of such a motion would not prevent the judgment from becoming final. Rule 60 provides that "a motion under this subdivision (b) does not affect the finality of a judgment or suspend its operation." Buggs v. Elgin, Joliet & Eastern R.R., 852 F.2d 318, 321 n. 3 (7th Cir.1988); Bernstein v. Lind-Waldock & Co., 738 F.2d 179, 182 (7th Cir.1984).

The second reason for believing that the affirmance and remand may have been a final decision after all is that the rules of finality that we have sketched are interpretations of 28 U.S.C. § 1291, which defines the appellate jurisdiction of the federal courts of appeals. The Supreme Court's appellate jurisdiction over state court decisions is defined by a different statute, 28 U.S.C. § 1257. See In re Riggsby, supra, 745 F.2d at 1156. Although the key word in the two statutes is the same--"final"--it has been interpreted differently because the relation between the Supreme Court and the state courts is different from the relation between the federal courts of appeals and the federal district courts. 15A Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure § 3908 (1992). The only rulings by a state court that the Supreme Court can review are rulings on issues of federal law, Illinois v. Gates, 462 U.S. 213, 217-18, 103 S. Ct. 2317, 2321, 76 L. Ed. 2d 527 (1983), whereas a federal court of appeals' appellate jurisdiction over rulings by district courts, the courts of the same sovereign, is plenary. Salve Regina College v. Russell, 499 U.S. 225, ----, 111 S. Ct. 1217, 1221, 113 L. Ed. 2d 190 (1991). This distinction has led the Court to deem a state-court judgment final if "the federal issue, finally decided by the highest court in the State, will survive and require decision regardless of the outcome of future state-court proceedings." Cox Broadcasting Corp. v. Cohn, 420 U.S. 469, 480, 95 S. Ct. 1029, 1038, 43 L. Ed. 2d 328 (1975); see also Brady v. Maryland, 373 U.S. 83, 85 and n. 1, 83 S. Ct. 1194, 1195 and n. 1, 10 L. Ed. 2d 215 (1963); American Export Lines, Inc. v. Alvez, 446 U.S. 274, 279 and n. 7, 100 S. Ct. 1673, 1676 and n. 7, 64 L. Ed. 2d 284 (1980); Radio Station WOW, Inc. v. Johnson, 326 U.S. 120, 65 S. Ct. 1475, 89 L. Ed. 2092 (1945). The remand in the present case was confined to a pure issue of state law--whether Richardson should receive a lighter sentence because evidence in a subsequent trial indicated that he had played a lesser role in the offense than the sentencing judge had reasonably believed to be the case. Even if Richardson got a lighter sentence on remand this would not moot his Batson issue, which concerned the validity of the conviction rather than the severity of the sentence. Thus the federal issue would survive and require decision regardless of the outcome of the further proceedings in the state court.

But does the Illinois Appellate Court's decision affirming Richardson's conviction fit the description in Cox of a final decision of the federal issue by the highest court in the state? Richardson did not attempt to appeal the intermediate court's decision to the Supreme Court of Illinois. One might be tempted to think that if, as a result, he could never raise the federal issue in that court, this would make the intermediate court's decision final within the meaning of Cox. Brown v. Texas, 443 U.S. 47, 50, 99 S. Ct. 2637, 2640, 61 L. Ed. 2d 357 (1979). One would be wrong. If Richardson could have sought leave to appeal to the state supreme court--and he could have, whether the decision of the intermediate court was final or not, because the supreme court has power to hear interlocutory appeals, Ill.Rev.Stat. ch. 110A, pp 315, 318(b)--then he could not, by declining to follow that route, make the intermediate court's decision final for purposes of obtaining review by the U.S. Supreme Court. Robert L. Stern, Eugene Gressman & Stephen M. Shapiro, Supreme Court Practice § 3.15 at p. 140 (6th ed. 1986); 16 Charles Alan Wright et al., Federal Practice and Procedure § 4007 at p. 554 (1977). And if after he was resentenced, and his sentence affirmed by the intermediate court, he could have obtained review of the federal issue by the state supreme court, the intermediate court's decision on his first appeal was on no account final within the meaning of Cox. As a matter of first principles, we might have expected that he could have obtained this delayed review by the state supreme court. It is easy to see why the U.S. Supreme Court, given its limited jurisdiction over state court judgments, might want to consider an otherwise interlocutory decision final because it disposed finally of all federal issues in the case; the Court would not have to worry about having to hear a second appeal in the same case after the proceedings on remand were completed. It is more difficult to see why a state supreme court should take such a position, as it would require that court to consider a defendant's appeal piecemeal. The court might be expected to prefer waiting until Richardson had finally been sentenced before considering whatever medley of federal and state grounds he advanced to upset the judgment.

No doubt in some cases it would make even more sense to hear and decide an appeal before the last possible moment, in order to head off time-consuming future proceedings. This is the rationale for allowing interlocutory appeals. But the Supreme Court of Illinois has, as we noted, express power to hear interlocutory appeals and therefore no need to jigger normal principles of finality in order to obtain jurisdiction over a case before the completion of proceedings on remand.

However all this may be, in Richardson's collateral proceeding the Illinois Appellate Court held (in an unpublished opinion) that the December 30, 1985, decision had indeed been a final decision so that Richardson's Batson issue could no longer be considered by the state court system unless he sought leave to appeal the decision to the state supreme court, which he did not. The discussion by the Illinois Appellate Court of the issue of finality is brief, but we are not disposed to quarrel with the conclusion. That the remand was in the nature of a collateral proceeding, like a remand to consider a motion for postjudgment relief, and therefore did not affect the finality of the December 30 decision, is entirely plausible and is supported (though not compelled) by two decisions of the state supreme court. People v. Holman, 132 Ill. 2d 128, 138 Ill. Dec. 155, 159, 547 N.E.2d 124, 128 (1989); People v. Williams, 404 Ill. 624, 89 N.E.2d 822 (1950).

It can be argued that even if the decision was final within the state court system, it might not be final for purposes of deciding whether Richardson is barred from invoking Batson. For it was not the very last decision from which Richardson could have appealed in order to get the question of the constitutional validity of his conviction before the U.S. Supreme Court. The Court could review a ruling in that decision by granting certiorari not only when the ruling was made with finality within the state court system, but also later, when the state supreme court denied Richardson leave to appeal from the second decision of the intermediate court (the one affirming the 30-year sentence imposed on remand from its first decision), even if no new federal issue was raised. Reece v. Georgia, 350 U.S. 85, 87, 76 S. Ct. 167, 169, 100 L. Ed. 77 (1955); Urie v. Thompson, 337 U.S. 163, 172-73, 69 S. Ct. 1018, 1026, 93 L. Ed. 1282 (1949). So even if the judgment in the first round of appeals was final, the judgment in the second round may also have been final, though there is no dispositive authority on the question. Stern, Gressman & Shapiro, supra, § 3.10 at pp. 132-33; 16 Wright et al., supra, § 4010 at pp. 598-99. We are not guilty of paradox in speaking of successive final judgments in the same case; such a succession is found in any case in which the first judgment is reversed and a second judgment is entered on remand. An interpretation of Cox that required an appellant to ask for certiorari at the earliest possible opportunity, at peril of losing his federal right, would merely flood the Supreme Court with protective appeals. 16 Wright et al., supra, § 4010 at p. 598.

But even if, as we believe quite likely, Richardson could wait to petition the Supreme Court for certiorari until his case was completely finished in the state courts, and could accumulate all his (nonwaived) federal claims for that petition, this would not avoid the decisions on retroactivity that we cited at the outset of this opinion. They hold that an applicant for federal habeas corpus cannot invoke a constitutional right that was first declared after his conviction became final and the time for seeking certiorari to review that final decision lapsed (or certiorari was denied). When that time ends or, a timely petition having been filed, certiorari is denied, the federal law applicable to the defendant's case is frozen; cases decided after that point cannot help him. The fact that events occurring in the state court system after his conviction has become final might entitle him to file another petition for certiorari later on does not detract from the finality of his conviction. Otherwise the happenstance of a remand that the Illinois Appellate Court has held did not affect the finality of Richardson's conviction would give him federal rights denied a similarly situated defendant whose conviction had been affirmed without a remand. Bear in mind that the event which kept Richardson's case alive after the affirmance of his conviction was a remand in the nature of a collateral proceeding. The essence of Teague and the other decisions on retroactivity in habeas corpus is that changes in the law that occur after the criminal proceeding has passed from the direct to the collateral phase are not a permissible basis for granting relief.

This point has been obscured in this case by the fact that when in the second round of his state court proceedings Richardson asked the state supreme court for leave to appeal from the intermediate court's affirmance of his new sentence, he neglected to raise the Batson issue, even though Batson had been decided two years earlier. This makes it look as if the problem in this case is waiver. And it is true that had he sought and obtained certiorari to review the affirmance of the new sentence the Supreme Court would not have reached the Batson question because it had not been presented to the state supreme court. 16 Wright et al., supra, § 4022 at pp. 696-97. But this bar to reversal on the basis of Batson would have had nothing to do with finality. It would have been the consequence of a routine forfeiture caused by a procedural bobble. The bobble could have been expected to forfeit Richardson's rights throughout the federal court system, since an issue not preserved within the state court system is deemed waived on federal habeas corpus, Smith v. Murray, 477 U.S. 527, 533, 106 S. Ct. 2661, 2665, 91 L. Ed. 2d 434 (1986), and if Richardson waived the Batson issue in his second round he could not urge the reversal of his conviction now on the basis of Batson even if that case had been decided fifty years ago. He escapes this fate only because, when he mounted a collateral attack on his conviction between the denial by the Illinois supreme court of leave to appeal from the second intermediate appellate decision and the filing of his application for federal habeas corpus, the state courts did not declare a forfeiture. They thought Richardson was barred from pleading Batson not by any procedural default that he had committed but simply because his conviction had become final before Batson had been decided. They were right. Richardson's conviction had become final before Batson had been decided; the fact that there was to be another final judgment in his case later did not revive his Batson claim.

AFFIRMED.

ILANA DIAMOND ROVNER, Circuit Judge, dissenting.

Under Griffith v. Kentucky, 479 U.S. 314, 107 S. Ct. 708, 93 L. Ed. 2d 649 (1987), Richardson is entitled to retroactive application of Batson if his conviction was not yet final when Batson was decided. For that purpose, his conviction was final if " 'the judgment of conviction was rendered, the availability of appeal exhausted, and the time for petition for certiorari had elapsed before [the] decision in' Batson v. Kentucky." Allen v. Hardy, 478 U.S. 255, 258 n. 1, 106 S. Ct. 2878, 2879 n. 1, 92 L. Ed. 2d 199 (1986) (per curiam) (quoting Linkletter v. Walker, 381 U.S. 618, 622 n. 5, 85 S. Ct. 1731, 1734 n. 5, 14 L. Ed. 2d 601 (1965)); see also Griffith, 479 U.S. at 321 n. 6, 107 S. Ct. at 712 n. 6; Teague v. Lane, 489 U.S. 288, 295, 109 S. Ct. 1060, 1067, 103 L. Ed. 2d 334 (1989).

This definition of finality reflects Griffith's underlying concern with treating similarly situated defendants similarly. In other words, when the Court happens to select one defendant's case "from the stream of appellate review" and announces a new rule of law, similarly situated cases that happen by mere chance to "flow by" should also benefit from that rule. 479 U.S. at 323, 107 S.Ct. at 713. Thus, if Richardson might have stood in Batson's stead had the Supreme Court selected his rather than Batson's case for review, then Richardson ought also to benefit from the newly-announced rule.

The majority recognizes that Richardson had properly preserved his Batson claim by raising it in pre-Batson form during his first round of state court review. Ante at 464. The majority also acknowledges that under Reece v. Georgia, 350 U.S. 85, 87, 76 S. Ct. 167, 169, 100 L. Ed. 77 (1955), and Urie v. Thompson, 337 U.S. 163, 172-73, 69 S. Ct. 1018, 1026, 93 L. Ed. 1282 (1949), the Supreme Court could have granted certiorari and heard Richardson's Batson claim after his second round of state court review, notwithstanding Richardson's failure to reassert the issue during the course of the latter review. Ante at 465. Thus, Richardson could have arrived at the Supreme Court with his Batson claim properly in tow after his second round of state court appeals. It seems clear, then, that his conviction was not final in Allen terms until that time, as the third prong of finality--expiration of the time for filing a certiorari petition--was satisfied only then. Richardson therefore was in the relevant Griffith stream when the Supreme Court decided Batson, and he, like Batson, should benefit from the new rule. Any other conclusion would violate Griffith 's technical requirements as well as its mandate of uniformity.

The majority avoids this result by characterizing Richardson's conviction as final after it was affirmed by the Illinois Appellate Court in December 1985 and the time to petition for certiorari review of that judgment had elapsed. The majority relies on a line of cases that examine the Supreme Court's certiorari jurisdiction over state court judgments. See ante at 466-67, citing Cox Broadcasting Corp. v. Cohn, 420 U.S. 469, 480, 95 S. Ct. 1029, 1038, 43 L. Ed. 2d 328 (1975); Brady v. Maryland, 373 U.S. 83, 85 & n. 1, 83 S. Ct. 1194, 1195 & n. 1, 10 L. Ed. 2d 215 (1963); American Export Lines, Inc. v. Alvez, 446 U.S. 274, 279 & n. 7, 100 S. Ct. 1673, 1676 & n. 7, 64 L. Ed. 2d 284 (1980); Radio Station WOW, Inc. v. Johnson, 326 U.S. 120, 65 S. Ct. 1475, 89 L. Ed. 2092 (1945).

28 U.S.C. § 1257, which establishes the Court's certiorari jurisdiction over state court judgments, provides:

(a) Final judgments or decrees rendered by the highest court of a State in which a decision could be had, may be reviewed by the Supreme Court by writ of certiorari ...

Thus, a state court judgment must be final before the Supreme Court may exercise its certiorari jurisdiction. That finality is, of course, distinct from Allen finality, which is accomplished only after the time to petition for certiorari review of the final state court judgment has elapsed. In the Cox line of cases, the Court examines the innermost boundaries of its certiorari jurisdiction, focusing on the earliest procedural moment that it might properly hear a case. The premise of the majority's analysis thus appears to be that if the Supreme Court could have exercised certiorari jurisdiction over Richardson's case after the original affirmance of his conviction, then Allen finality occurred when the time for petitioning for certiorari review of that judgment expired.

But reliance on the Cox line of cases is misplaced in the retroactivity context. First, contrary to the majority's characterization, those cases do not actually speak in terms of "finality" of the state court judgment. Instead, they ask when the Supreme Court might exercise certiorari jurisdiction before strict finality of the state court judgment. The cases have thus considered the rules for early review to be exceptions to the finality requirement. Cox, for example, frames its discussion as follows:

[T]he Court [in Radio Station WOW v. Johnson ] observed that the [finality] rule had not been administered in such a mechanical fashion and that there were circumstances in which there has been "a departure from this requirement of finality for federal appellate jurisdiction."

* * * * * *

There are now at least four categories of such cases in which the Court has treated the decision on the federal issue as a final judgment for the purposes of 28 U.S.C. § 1257 and has taken jurisdiction without awaiting the completion of the additional proceedings anticipated in the lower state courts.

420 U.S. at 477, 95 S. Ct. at 1037.1 American Export Lines, Inc. v. Alvez, 446 U.S. 274, 277, 100 S. Ct. 1673, 1675, 64 L. Ed. 2d 284 (1980), similarly describes the Cox rules as "categorical exceptions to strict finality which the Court has developed in construing § 1257." More recently, Pennsylvania v. Ritchie, 480 U.S. 39, 50 n. 8, 107 S. Ct. 989, 997 n. 8, 94 L. Ed. 2d 40 (1987), has again referred to the Cox categories as "exceptions" to the finality doctrine. See also id. at 73, 107 S.Ct. at 1009 (Stevens, J., dissenting) ("In Cox ... the Court recognized some limited exceptions to the general principle that this Court may not review cases in which further proceedings are anticipated in the state courts." (citation omitted)).

Contrary to the majority's description, then, the Court does not envision that state court finality is actually accomplished more than once in any given case. The point of the cases is not that the state court judgment is final at the earlier time, but only that it may then be reviewable on certiorari. Thus, whether or not the Illinois Appellate Court's original ruling might come within one of the Cox exceptions sheds no light on the "finality" of the state court judgment or, in turn, on Allen finality, which would follow thereafter.

Even more fundamentally, there is no indication that the Cox exceptions include a restrictive component. That is, the cases do not suggest that an appellant who might have qualified for early review loses the opportunity for review by not petitioning for certiorari until after strict finality has been accomplished. Granted, the Supreme Court has not spoken directly to this issue. As the majority points out, however, such a reading of the cases is as undesirable as it is improbable. Ante at 467. Nor is there any reason to find Supreme Court silence on the issue to be meaningless. The issue has presumably never been discussed precisely because the Court does not engage in a Cox inquiry when cases come to it as section 1257 envisions, after strict finality of the state court judgment. The propriety of review at that time is presumably not in question.2

Yet if the cases do not foreclose later certiorari review, as the majority seemingly agrees they do not, then they are not relevant to the retroactivity discussion. If, despite the fact that the Supreme Court may have accepted the case earlier, an appellant may also petition for certiorari later, then whether or not the Court might have granted early review is not relevant to whether the time to petition for certiorari has elapsed. That will of course not occur until the latest time that a certiorari petition could have been filed. As long as Richardson could have petitioned for certiorari and the Court could have reviewed the federal issues decided in the earlier state court appeal, Richardson was in the stream envisioned by Griffith and could himself have reached the Supreme Court on his Batson claim had the Court not by chance selected Batson's case instead.

Moreover, the fact that the Illinois Appellate Court has itself characterized Richardson's conviction as final at the earlier time should have no bearing on our analysis.3 See ante at 467. In discussing finality for purposes of certiorari jurisdiction, Cox has explicitly rejected reliance on state court characterizations of finality:[W]ith respect to finality: "The designation given the judgment by state practice is not controlling. The question is whether it can be said that 'there is nothing more to be decided' that there has been 'an effective determination of the litigation.' ..."

420 U.S. at 479 n. 8, 95 S. Ct. at 1038 n. 8 (quoting Richfield Oil Corp. v. State Board, 329 U.S. 69, 72, 67 S. Ct. 156, 158, 91 L. Ed. 80 (1946)) (citations omitted).

Under Allen and Griffith, which articulate the only definition of finality that is relevant for retroactivity purposes, a conviction is final only after the time to petition for certiorari has elapsed. If a state court's determination that a conviction is final does not affect the Supreme Court's certiorari jurisdiction, then neither can it affect the "finality" of the conviction for purposes of retroactive application of a rule of federal law. Thus, the Illinois Appellate Court's determination that Richardson's conviction had become final at the earlier time is not relevant here.

Finally, as a policy matter, the majority's approach has the precise effect that the majority concedes ought to be avoided--that of encouraging protective early certiorari petitions. See ante at 467. Batson was decided only four months after the Illinois Appellate Court had affirmed Richardson's conviction on the first round of review. If Richardson had pursued his Batson claim through the highest state court and petitioned for certiorari at that time, his review presumably would not have been completed before the Supreme Court decided Batson. Thus, once the Batson certiorari petition had been granted, effective representation of a defendant in Richardson's position would require the filing of an early certiorari petition in order to prolong the review process sufficiently to qualify for retroactive application of the potential new rule.4

Richardson properly preserved his Batson claim, and, as the majority acknowledges, he could have reached the Supreme Court with that issue after the second round of state court review. His conviction therefore was not "final" in Griffith and Allen terms before Batson was decided. To my mind, the majority has offered no compelling reasons for straying from that straightforward result. Instead, it has contradicted Griffith's clear mandate of uniformity by refusing Richardson the retroactive benefit of Batson.

For these reasons, I respectfully dissent.