CUDAHY, Circuit Judge. William Kelly was part of a large drug conspiracy. He and 13 other defendants were charged with a variety of offenses relating to a scheme to import and distribute large quantities of drugs. According to the presentence report, Kelly was responsible for importing and/or distributing 300 kilograms of cocaine and more than seven tons of marijuana.
He was convicted and sentenced to 35 years in prison. Kelly appealed his conviction and sentence, claiming a variety of errors. In a previous opinion we affirmed. See United States v. Pallais, 921 F.2d 684 (7th Cir. 1990), cert. denied, 116 L. Ed. 2d 101, 112 S. Ct. 134 (1991).
Kelly then filed a motion under 28 U.S.C. § 2255, challenging his sentence. In particular, he argued that the district court did not make specific findings of fact as Fed. R. Crim. P. 32(c)(3)(D) requires. He also claimed that his sentence could not lawfully be enhanced by virtue of his previous drug conviction, because the government had not served either him or his lawyer with timely notice of that conviction under 21 U.S.C. § 851(a). The district court refused to hear this motion, concluding that his failure to raise these issues earlier represented a procedural default, and that Kelly failed to make a sufficient showing of "cause and prejudice" to overcome that default. Kelly appeals, and we reverse.
Kelly was convicted on 12 counts. He was sentenced to concurrent sentences on each of them, and as a result faced a 35-year term of imprisonment. At issue here are counts 1, 25 and 27. Count 1 charged that Kelly participated in a drug conspiracy in violation of 21 U.S.C. § 846. The maximum statutory penalty for violating this provision was 20 years imprisonment (the statute indicated no minimum). But under the statute prior drug offenders could have their sentences enhanced to up to 40 years. 21 U.S.C. § 841(b)(1)(A) (indicating 1986 amendments). Kelly had a previous conviction (for a drug offense in New Jersey in 1979) which was used to enhance his sentence. The court therefore sentenced him on this count to 25 years.
Counts 25 and 27 both alleged that Kelly imported marijuana into the United States, in violation of 21 U.S.C. § 952(a). Under 21 U.S.C. § 960(b), those who violate § 952 by importing large quantities of drugs are to be sentenced to terms of imprisonment of 5-40 years. 21 U.S.C. § 960(b)(2). But if the defendant has a previous drug offense, the range is enhanced to 10 years to life imprisonment. Id. The district court therefore originally sentenced Kelly to a 60-year prison term, but later (on examining the sentences imposed on Kelly's co-defendants) reduced the sentence to 35 years.
Kelly's primary argument is that the previous conviction cannot be used to enhance his sentence because neither he nor his lawyer was served with notice of the conviction before trial.1 Section 851 serves a variety of purposes. First, § 851(a) says that if a defendant is to receive an enhanced penalty because of her previous convictions, the government must tell the court and the defendant about the convictions on which it intends to rely before trial begins. Second, if the government has complied with § 851(a), § 851(b) requires the court, after conviction but before sentencing, to ask the defendant whether she admits or denies that she has previously been convicted, and to tell her that, if she does not challenge the previous conviction before sentencing, she cannot later object to the sentence on that basis. But § 851(e) provides that, if the conviction is more than five years old, the defendant cannot challenge it. This court has therefore recently held that, where this is the case, the district court need not comply with the ritual otherwise required by § 851(b). United States v. Flores, 5 F.3d 1070 (7th Cir. 1993), cert. denied, 127 L. Ed. 2d 79, 114 S. Ct. 884 (1994).
But the reasoning of Flores does not extend to § 851(a). The purpose of § 851(b) is to give the defendant an opportunity to challenge a previous conviction. Section 851(a) serves a different and additional purpose, to inform the defendant that she faces severe consequences if convicted. This procedure, one would hope, should lead to better informed decisions whether to proceed to trial. But because a defendant's sentence can be enhanced no matter how old the previous conviction, the relief from the strictures of § 851(b) that Flores provides does not extend to § 851(a). So regardless of the previous conviction's vintage, § 851(a) requires that the defendant be served with notice if it is to be used to enhance her sentence. In fact, we have previously observed that a court does not even have jurisdiction to impose an enhanced sentence unless notice is served. United States v. Belanger, 970 F.2d 416, 418 (7th Cir. 1992) ("Failure to file the notice prior to trial deprives the district court of jurisdiction to impose an enhanced sentence."). This Conclusion presumably follows from the language of § 851(a), which directs that "no person who stands convicted of an offense under this part shall be sentenced to increased punishment by reason of one or more prior convictions" unless served with notice.
Here there is a dispute about whether Kelly received timely notice, even though the court ultimately found that the defendant (or more precisely, his lawyer) was served with notice prior to the swearing of the jury. The district court thought that that notice was good enough, and decided that it was "not going to make any further record on that unless the Court of Appeals decides that that's an important enough issue that they want more of a record developed on it." Sent. Tr. (Jan. 31, 1989) 18. It is; we do.
The question here is defining what § 851(a) means when it requires service "before trial." As far as we can tell every court of appeals to have addressed this question has concluded that before trial means before jury selection begins (which is obviously also before the jury is sworn). See United States v. Gonzalez-Lerma, 14 F.3d 1479, 1484 (10th Cir. 1994) ("the circuits that have considered this issue in the context of jury trials have concluded that filing anytime before jury selection begins is sufficient for purposes of 21 U.S.C. 851"); United States v. White, 980 F.2d 836, 842 (2d Cir. 1992) ("we conclude the phrase 'before trial' in § 851(a)(1) means before the commencement of jury selection"); United States v. Johnson, 944 F.2d 396, 407 (8th Cir.), cert. denied, 116 L. Ed. 2d 663, 112 S. Ct. 646 (1991) ("section 851 requires filing before jury selection begins"); United States v. Weaver, 905 F.2d 1466, 1481 (11th Cir. 1990), cert. denied, 498 U.S. 1091 (1991) (service timely because made "prior to the selection of the trial jury"); United States v. Jordan, 258 U.S. App. D.C. 143, 810 F.2d 262, 268-69 (D.C. Cir.), cert. denied, 481 U.S. 1032, 95 L. Ed. 2d 535, 107 S. Ct. 1963 (1987) (service timely when made "before voir dire and empaneling of the jury").
The government does not here ask us to split with our sister circuits, but rather invites us, in effect, to mis-read the record. We are urged to infer from the district court's conditional suggestion--"if I had to make a finding on that point I'd be more inclined to believe the attorneys for the government"--that the court actually found that service was made before jury selection began. We respectfully decline. The district court expressly said that it takes "the position that if filed prior to the swearing of the jury it's properly considered," and avowedly decided not to find that Kelly was served before jury selection began. "Exactly when I'm not positive," the court frankly admitted. Sent. Tr. (Jan. 31, 1989) at 18. This record does not permit us to conclude that the district court found that Kelly (or his lawyer) was served with notice before jury selection began.
As for the definition of the beginning of trial, we join the five other circuits to have examined this question and conclude that "before trial" means before the commencement of jury selection. Because the district court did not find that either Kelly or his attorney were timely served, 21 U.S.C. § 851 deprives the district court of jurisdiction to impose an enhanced sentence by virtue of Kelly's previous conviction.
But the government contends that this does not matter, because the district court never imposed an enhanced sentence anyway. On Counts 25 and 27, the court imposed a 35 year sentence. If, as the court believed, Kelly could receive an enhanced sentence, the statutory range would have been 10 years to life imprisonment. But the range that was in fact applicable (because of the failure to serve Kelly with notice) was 5 to 40 years.
The government here reasons that because the court might have come up with 35 years even had it gotten the law right, no harm has befallen Kelly by virtue of the court's mistake. We cannot accept this argument because it asks us to ignore the Supreme Court's insistence that "a litigant's failure to clear a jurisdictional hurdle can never be harmless," Torres v. Oakland Scavenger Co., 487 U.S. 312, 317 n.3, 101 L. Ed. 2d 285, 108 S. Ct. 2405 (1988). See also Weaver, 905 F.2d at 1481 (quoting United States v. Olson, 716 F.2d 850, 852 (11th Cir. 1983)) ("Significantly, 'the doctrine of harmless error does not apply' with respect to failures to follow the statutory scheme of § 851.").
And even if the application of the wrong range could be treated as harmless, we would be faced with the more general question of what to do where a court applies an incorrect sentencing range, but comes up with a number that (perhaps by coincidence) is inside the correct range. At least in the Guidelines context, the Supreme Court has indicated that the party challenging a sentence on appeal "does not have the . . . burden of proving that [the application of an incorrect range] was determinative in the sentencing decision." Williams v. United States, 117 L. Ed. 2d 341, 112 S. Ct. 1112, 1120 (1992). "Rather, once the court of appeals has decided that the district court misapplied the Guidelines, a remand is appropriate unless the reviewing court concludes, on the record as a whole, that the error was harmless, i.e., that the error did not affect the district court's selection of the sentence imposed." Id. at 1120-21. While this case involves a statutory rather than a Guidelines range, the rule of Williams appears to be a straight-forward application of the general harmless error rule embodied in Fed. R. Crim. P. 52(a). We therefore find the same analysis to be applicable.
Occasions in which this type of error are harmless seem to be relatively rare. The district court thought that it should pick a number between 10 and life. It came up with 35. What would have happened if (as it should have) it were choosing between 5 and 40? Could it still have been 35? Of course, but we cannot declare an error to be harmless just because, in a correct analysis, it is a mathematical possibility. Of course, the district court may indicate on the record that it would impose a particular sentence even if it were applying a different range. The government insisted at oral argument that the district court's adjustment of Kelly's sentence, designed to bring it into line with the sentences imposed on his co-defendants, puts this case into that category. But absent an express statement that the court would impose the same sentence even if a different range were applicable, it is difficult to imagine a case in which an appeals court could declare with the requisite degree of confidence that the application of an incorrect range would amount to a harmless error; we do not believe this to be such a case. Compare United States v. Mount, 966 F.2d 262, 265 (7th Cir. 1992) ("An appellate court may be confident that the choice of range did not affect the sentence when the district Judge says so.") (emphasis added). Thus, even if jurisdictional defects could be treated as harmless, we would not find this error to be a harmless one.
As for Count 1, the failure to timely serve Kelly means that the statutory maximum was 20 years. The court imposed a 25 year term of imprisonment on this count. The government again contends that this is of no moment, because the concurrent 35-year sentences are "controlling." But because "erroneous concurrent sentences are correctable," Wright v. United States, 519 F.2d 13, 17 (7th Cir.), cert. denied, 423 U.S. 932, 46 L. Ed. 2d 262, 96 S. Ct. 285 (1975), as long as Kelly is permitted to make his argument under § 851(a) on collateral review (a question we take up presently), on re-sentencing the district court should impose a sentence on Count 1 no greater than the statutory maximum.
The heart of the government's argument is that even if Kelly is right about all of this, he has waived this argument by not presenting it earlier. More precisely (since there is a technical difference between "waiver" and "procedural default"), the government insists Kelly's argument under § 851(a) is not cognizable on a § 2255 motion absent a showing of "cause and prejudice," a showing the government insists that Kelly cannot make.
It is true that Kelly did not make his § 851(a) argument either at trial or on direct appeal. Rather, he first advanced this argument when he filed his motion under § 2255. The district court therefore refused to hear the argument, finding it to have been procedurally defaulted. Section 2255 is the federal prisoner's analog to the petition for habeas corpus under 28 U.S.C. § 2254. But unlike the habeas petitioner, a prisoner seeking collateral review under § 2255 files a motion in the same court that originally convicted him, and that motion is considered part of the same case as his original conviction. See United States v. Frady, 456 U.S. 152, 182-83, 71 L. Ed. 2d 816, 102 S. Ct. 1584 & nn.5 & 6 (1982) (Brennan, J., Dissenting).
Relief under § 2255 is available only for errors of constitutional or jurisdictional magnitude, or where the error represents a "fundamental defect which inherently results in a complete miscarriage of Justice." Belford v. United States, 975 F.2d 310, 313 (7th Cir. 1992). In Frady, moreover, the Supreme Court held that the procedural default rules that apply in habeas cases following Wainwright v. Sykes, 433 U.S. 72, 53 L. Ed. 2d 594, 97 S. Ct. 2497 (1977), apply equally in § 2255 cases. Under that standard, an argument that was not presented on direct appeal cannot first be advanced on collateral review "absent a showing of cause" for the failure to advance the argument sooner "and some showing of actual prejudice resulting from the alleged constitutional violation." Id. at 84. See also Belford, 975 F.2d at 313 (applying the cause and prejudice standard to § 2255 and collecting cases).
The district court here said that because there was an insufficient showing of cause and prejudice (actually, prejudice is clear enough, the only real question centers on cause), Kelly could not advance his § 851(a) argument on collateral review. On appeal, Kelly attempts to rebut the district court on its own terms, insisting that he received ineffective assistance of counsel, and that this ineffectiveness constitutes cause and prejudice. It is true that ineffective assistance can constitute sufficient cause for the failure to raise an issue. Murray v. Carrier, 477 U.S. 478, 488, 91 L. Ed. 2d 397, 106 S. Ct. 2639 (1986) (saying so for the purposes of § 2254); Belford, 975 F.2d at 310 (applying Murray to § 2255). Kelly's argument here is essentially that his appellate counsel was ineffective for failing to raise this issue on direct appeal. We examined the standard for such a claim in Gray v. Greer, 800 F.2d 644 (7th Cir. 1985), and held that appellate counsel is ineffective if counsel fails to raise issues that are (1) obvious, and (2) clearly stronger than the ones raised. The government here notes that Kelly's appellate counsel raised a number of alleged defects on direct appeal which, although ultimately unsuccessful, this court addressed in a fairly lengthy published opinion in United States v. Pallais, 921 F.2d at 684.
But it seems to us that this whole argument misses the point. Frady is essentially a case about waiver. Arguments not raised at trial or on direct appeal are waived. If there are errors that were not objected to, Frady says that appellate courts are not to apply the "plain error" standard of Fed. R. Crim. P. 52(b), but can only grant relief if there is a showing of cause and prejudice. The point of cause and prejudice, we repeat, is to overcome the waiver. But this analysis of course assumes that the error in question is a waivable one. And jurisdictional defects are not.
Federal courts can exercise jurisdiction only where it is given to them by Congress. We suggested in Belanger, 970 F.2d at 418, that failure to comply with § 851 "deprives the district court of jurisdiction to impose an enhanced sentence." Other courts of appeals--and the government in this case--have also recognized compliance with § 851 to be a jurisdictional prerequisite. See United States v. Wright, 932 F.2d 868, 882 (10th Cir.), cert. denied, 116 L. Ed. 2d 448, 112 S. Ct. 428 (1991); United States v. Olson, 716 F.2d 850, 852-53 (11th Cir. 1983) (failure to comply with § 851 "restricts the court's authority to impose enhanced sentences"); Government's Br. at 6 ("Section 851 is of jurisdictional, not constitutional consequence.").
This Conclusion carries significant consequences since for centuries it has been recognized that federal courts have an obligation--regardless of the arguments advanced to them by the parties--to assure themselves of their own jurisdiction. See Capron v. Van Noorden, 6 U.S. (2 Cranch) 126, 127, 2 L. Ed. 229 (1804) ("Here it was the duty of the court to see that they had jurisdiction, for the consent of the parties could not give it."); Louisville & Nashville R. Co. v. Mottley, 211 U.S. 149, 152, 53 L. Ed. 126, 29 S. Ct. 42 (1908) ("Neither party has questioned that jurisdiction, but it is the duty of this court to see to it that the jurisdiction of the circuit court, which is defined and limited by statute, is not exceeded. This duty we have frequently performed on our own motion."). See also Commercial Nat'l Bank of Chicago v. Demos, No. 92-3839, slip op. at 4 (7th Cir. Mar. 10, 1994) ("That the parties have not contested, nor the district court considered jurisdiction does not impede our inquiry . . . We must consider the issue sua sponte when it appears from the record that jurisdiction is lacking.").
The implication of all of this is that questions about the court's jurisdiction cannot be waived. See Freytag v. Commissioner of Internal Revenue, 111 S. Ct. 2631, 2648 (1991) (Scalia, J., Concurring) (discussing the "'nonwaivability' of lack of subject-matter jurisdiction"). As noted, the court has an independent duty to assure itself that its jurisdiction is properly had, and as a result parties can raise jurisdictional defects at any time. The only limitation on this principle is that res judicata attaches to a court's finding that it has jurisdiction, see Insurance Corp. of Ireland, Ltd. v. Compagnie des Bauxites de Guinee, 456 U.S. 694, 702 n.9, 72 L. Ed. 2d 492, 102 S. Ct. 2099 (1982), so a party cannot in a separate proceeding object to the enforcement of an adverse judgment on jurisdictional grounds. Treinies v. Sunshine Mining Co., 308 U.S. 66, 78, 84 L. Ed. 85, 60 S. Ct. 44 (1939) (quoting American Surety Co. v. Baldwin, 287 U.S. 156, 166, 77 L. Ed. 231, 53 S. Ct. 98 (1932)) ("'The principles of res judicata apply to questions of jurisdiction as well as to other issues,' as well as to jurisdiction of the subject matter as of the parties."). But the doctrine of res judicata of course does not apply either to state prisoner habeas petitions, see Brown v. Allen, 344 U.S. 443, 458, 97 L. Ed. 469, 73 S. Ct. 397 (1953), or to § 2255 motions, e.g., United States v. Donn, 661 F.2d 820 (9th Cir. 1981); so there is no barrier to a court's examination, on a § 2255 motion, of the propriety of its original exercise of jurisdiction. Rather a court is under an affirmative obligation to so inquire.2
Because a jurisdictional defect cannot be procedurally defaulted in the first place, it is therefore a non sequitur to suggest that a procedural default cannot be overcome because the defendant has not made an adequate showing of "cause." Kelly, somewhat to our surprise, never explicitly makes this argument, contending all the while that he made a sufficient showing of cause and prejudice. But because we are here dealing with a jurisdictional question, we heed the Supreme Court's admonition that "every federal appellate court has a special obligation to satisfy itself not only of its own jurisdiction, but also that of the lower courts in a cause under review." Bender v. Williamsport Area School Dist., 475 U.S. 534, 541, 89 L. Ed. 2d 501, 106 S. Ct. 1326 (1986). And the very point of our analysis is that the district court lacked jurisdiction to impose an enhanced sentence. Kelly's § 2255 motion was part of the same case, and it is therefore our obligation--despite Kelly's possible concession that the issue is not before us unless we find cause and prejudice--to vacate the sentence imposed without jurisdiction.
We also find it somewhat surprising that we have been unable to find a court that has previously held Frady's "cause and prejudice" requirement to be inapplicable where the alleged defect is jurisdictional. The First Circuit has once suggested that this may be the case, see Hardy v. United States, 691 F.2d 39, 41 (1st Cir. 1982) ("While it is not clear that Frady applies where a court acts without authority, as opposed to erring within the range of its authority, the existence of prejudice is clear.") And in an unpublished opinion the Ninth Circuit suggested that "because [the error] is jurisdictional, [the defendant] need not show cause and prejudice." United States v. Broadwell, 1992 U.S. App. Lexis 6366, *5 (9th Cir.). See also United States v. Milestone, 626 F.2d 264, 267 n.2 (3d Cir.), cert. denied, 449 U.S. 920, 66 L. Ed. 2d 148, 101 S. Ct. 319 (1980) (observing, before Frady, that "although petitioner did not raise this issue on direct appeal, collateral review is available under § 2255 to correct jurisdictional errors"); Thor v. United States, 554 F.2d 759, 762 (5th Cir. 1977) ("Jurisdictional defects are always subject to attack under section 2255, as that statute expressly states."). But despite the absence of more affirmative supporting authority, the logic of the analysis is clear enough. Because jurisdictional defects are nonwaivable, Kelly need not provide us with an excuse ("cause and prejudice") adequate to convince us to forgive his waiver. That being the case, the judgment of the district court, denying Kelly's motion for relief under § 2255, must be reversed.
Kelly also contends that he made several objections to alleged factual inaccuracies in the presentence report, and that the district court did not make a sufficient written record of its findings as to the controverted matter to satisfy Fed. R. Crim. P. 32(c)(3)(D). In light of our analysis above necessitating that the matter be remanded to the district court for resentencing, these objections are moot, though it does not appear to us that a failure to make a required factual finding would represent a constitutional or jurisdictional error, or a "fundamental defect which inherently results in a complete miscarriage of Justice," Belford, 975 F.2d at 313. Such errors (unless the failure to find facts rises to the level of a due process violation) are therefore not properly cognizable in a § 2255 motion. But as long as this case is in any event going back to the district court for resentencing, we urge that court to pay careful attention to the requirements imposed upon it by Fed. R. Crim. P. 32(c)(3)(D).
The dismissal of Kelly's § 2255 motion is REVERSED and his sentence VACATED. The matter is REMANDED to the district court for resentencing.
* Hon. Ann Claire Williams of the Northern District of Illinois is sitting by designation.
1. 21 U.S.C. § 851(a)(1) says that "no person who stands convicted of an offense under this part shall be sentenced to increased punishment by reason of one or more prior convictions, unless before trial, or before entry of a plea of guilty, the United States attorney files an information with the court (and serves a copy of such information on the person or counsel for the person) stating in writing the previous convictions to be relied upon."
2. The original purpose of the writ of habeas corpus, of course, was to allow relief where a defendant was convicted by a court that lacked jurisdiction. See, e.g., Johnson v. Zerbst, 304 U.S. 458, 82 L. Ed. 1461, 58 S. Ct. 1019 (1938).