These cross-appeals require us to explore poorly understood terrain: the modern configuration and use of the ancient writ of error coram nobis. This archeological dig entails a determination of when and under what circumstances that writ may be used by a criminal defendant as a vehicle for securing post-conviction relief.
The context is as follows. The petitioner, Mfred W. Trenkler, is a prison inmate, currently serving a federal sentence. He confronted the district court with a claim that more than ten years earlier it had illegally sentenced him to life imprisonment. Persuaded by this claim, the district court granted a writ of error coram nobis, vacated the original sentence, and proceeded to resentence the petitioner to a term of years. The government protests this deployment of the writ, contending that the district court lacked jurisdiction to issue it. The petitioner cross-appeals, *89maintaining that even the substituted sentence exceeded the maximum available under the statutes of conviction.
After first grappling with a series of threshold challenges to our appellate jurisdiction, we conclude that the district court had no authority to issue a writ of error coram nobis. Consequently, we reverse the order granting the writ and direct the district court to reinstate the original sentence.1
I. BACKGROUND
We rehearse here only those facts that are needed to place these cross-appeals into perspective. We urge the reader who hungers for more exegetic detail to consult our earlier opinions describing the various way stations that dot the trail of the petitioner’s case. See, e.g., Trenkler v. United States (Trenkler III), 268 F.3d 16 (1st Cir.2001) (affirming denial of relief under 28 U.S.C. § 2255); United States v. Trenkler (Trenkler II), No. 97-1239, 1998 WL 10265 (1st Cir. Jan.6,1998) (affirming denial of Rule 33 motion); United States v. Trenkler (Trenkler I), 61 F.3d 45 (1st Cir.1995) (affirming conviction and sentence on direct review); see also Trenkler v. Pugh (Trenkler IV), 83 Fed.Appx. 468 (3d Cir.2003) (upholding denial of relief under 28 U.S.C. § 2241).
On November 29, 1993, a jury convicted the petitioner of conspiracy, 18 U.S.C. § 371, and the illegal receipt and use of explosives, id. §§ 844(d), 844(i). The charges arose out of the petitioner’s role in a bombing that caused the death of one Boston police officer and the maiming of another. See Trenkler I, 61 F.3d at 47-48. The district court sentenced the petitioner to 60 months’ imprisonment on the conspiracy count and life imprisonment on each of the two “explosives” counts. All of the sentences were to run concurrently.
On the date that the indictment was handed up, both sections 844(d) and 844(i) provided in pertinent part that
if death results to any person, including any public safety officer performing duties as a direct or proximate result of conduct prohibited by this subsection, [the defendant] shall [also] be subject to imprisonment for any term of years, or to the death penalty or to life imprisonment as provided in section 34 of this title.
Section 34 provided in turn that when any of the proscribed acts “resulted in the death of any person,” the death penalty or life imprisonment would be available as a punishment “if the jury shall in its discretion so direct.” Notwithstanding this language, no special jury finding was either requested or obtained in the petitioner’s case. Indeed, the anomaly went unremarked.
Following the imposition of sentence, the petitioner appealed. His appeal challenged a number of evidentiary rulings. See Trenkler I, 61 F.3d at 51-62. We rejected all but one of these challenges, found that the successful challenge embodied a harmless error, and affirmed the judgment below. Id.
The next ten years witnessed a kaleidoscopic array of post-conviction proceedings. On December 22, 1995, the petitioner moved for a new trial or in the alternative an evidentiary hearing on the ground of newly discovered evidence. Agreeing that the proffered evidence was not newly discovered, we affirmed the *90district court’s denial of the motion. Trenkler II, 1998 WL 10265, at *4. The district court and this court subsequently rejected as time-barred an application, brought under 28 U.S.C. § 2255, that alleged ineffective assistance of counsel. See Trenkler III, 268 F.3d at 27. The petitioner’s other efforts met a similar fate. See, e.g., Trenkler IV, 83 Fed. Appx. at 472.
On August 24, 2004, the petitioner sought a writ of mandamus in this court, claiming for the first time that 18 U.S.C. §§ 844(d) and 844(i) required specific jury authorization as a condition precedent to the imposition of a life sentence. We summarily rejected his foray, noting that it amounted to a second or successive section 2255 petition and, as such, ran afoul of the gatekeeping provisions of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), Pub.L. No. 104-132, 110 Stat. 1214.
Undeterred by this setback, the petitioner wrote to the original trial judge, making essentially the same argument. The district court appointed counsel and, on November 6, 2006, appointed counsel moved for the issuance of a writ of error coram nobis or in the alternative a writ of audita querela.2 Although the petitioner used the docket number of the original criminal case, the district court treated the filing as constituting a new civil action and assigned a new docket number to it. The court reasoned that coram nobis, like its habeas corpus counterpart, is civil in nature.
The court ordered the government to respond by January 5, 2007. That date came and went without any word from the government. Acting sua sponte, the court extended the time for filing an opposition to February 2, 2007. It simultaneously warned that, should no opposition be filed within the extended period, it would decide the matter without the government’s participation. That deadline, too, expired without any action by the government.
On February 20, 2007, the district court granted a writ of error coram nobis. Trenkler v. United States (Trenkler V), No. 06-12072, 2007 WL 551620, at *1 (D.Mass. Feb.20, 2007). The court asserted jurisdiction under the All Writs Act, 28 U.S.C. § 1651, and expressed the view that the writ was available in criminal proceedings to correct fundamental errors, as long as other remedies were unavailable. Trenkler V, 2007 WL 551620, at *3. It then explained why it thought that the petitioner’s claim warranted such extraordinary relief.
To begin, the court concluded that the express language of the statutes of conviction required a jury directive as a condition precedent to the imposition of a life sentence. Id. at *1. It noted that the petitioner’s life sentence had been imposed in the absence of such a directive and was, therefore, illegal. Id This bevue warranted coram nobis relief because it was “fundamental to the validity of the judgment,” id at *4, and coram nobis represented the only procedural mechanism that remained *91available to correct the sentence, id. at *2. In reaching this conclusion, the court conceded that the time had expired for filing a section 2255 petition, see 28 U.S.C. § 2255(f), but suggested that the petitioner could not be faulted for failing to seek relief earlier since all concerned — the court, the government, and the petitioner’s previous counsel — had been caught flatfooted by this sentencing limitation. See Trenkler V, 2007 WL 551620, at *5.
When all was said and done, the court vacated the petitioner’s two concurrent life sentences and set the criminal case down for resentencing on April 4, 2007. The day after the court’s order issued, the government moved for leave to file an opposition to the coram nobis petition, blaming its failure to respond in a timely manner on “institutional inadvertence.” The district court granted the government’s request. The government not only filed an opposition but also moved for reconsideration of the coram nobis order. See Fed.R.Civ.P. 59(e). In subsequent filings, it asserted among other things that the district court lacked jurisdiction to issue the writ because the petitioner’s request was merely a camouflaged section 2255 petition and, as such, was subject to AEDPA’s limitations on second and successive habeas petitions. It also claimed that coram nobis relief was inappropriate because the petitioner was still in custody, because he had failed to establish a fundamental error in the criminal proceedings, and because he had failed to prove any good cause for not raising the alleged sentencing error at an earlier time.
On April 4, 2007, the district court, ruling from the bench, dismissed the government’s objections and reaffirmed the appropriateness of coram nobis relief. When the district court proceeded to the criminal case, the petitioner argued that the doctrine of Apprendi v. New Jersey, 530 U.S. 466, 490, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000), limited the length of an incarcera-tive sentence on the affected counts to ten years. The court rejected that argument and imposed concurrent sentences of 37 years’ immurement on the two affected counts (crediting the petitioner with time served to that point).
On April 9, 2007, the court entered an amended judgment in the criminal case, memorializing the new sentence. Three days later, the court entered a final judgment in the separate coram nobis proceeding, memorializing the grant of the writ and the annulment of the original sentence. On April 17, the government appealed from the final judgment in the co-ram nobis proceeding and, the next day, appealed from the amended final judgment in the criminal case. On April 26, the petitioner filed notices of appeal in both the civil and criminal cases.3 Although these appeals were consolidated for briefing and argument in this court, we need look no further than the cross-appeals in the coram nobis proceeding. See supra note 1.
II. ANALYSIS
The focal point of our analysis is the government’s assertion that the district court lacked authority to issue a writ of error coram nobis. There is, however, a logically antecedent issue: the petitioner’s suggestion that we have no jurisdiction to review the order granting coram nobis relief. We start there.
*92A. Appellate Jurisdiction.
The judgment in the coram nobis proceeding, on its face, seems final and ap-pealable. It leaves nothing to be done but the execution of the judgment itself. Catlin v. United States, 324 U.S. 229, 233, 65 S.Ct. 631, 89 L.Ed. 911 (1945); United States v. Metro. Dist. Comm’n, 847 F.2d 12, 14 (1st Cir.1988). The petitioner demurs. In this regard, his most loudly bruited claim is that a grant of coram nobis constitutes an unreviewable act of judicial discretion.
The petitioner’s sole support for this proposition is fragmentary language from an obscure Supreme Court opinion, Pickett’s Heirs v. Legerwood, 32 U.S. (7 Pet.) 144, 148, 8 L.Ed. 638 (1833). That case, which dates back to the early nineteenth century, involved a trial court’s issuance of what amounted to a writ of error coram nobis4 in an action for trespass and ejectment. The Justices concluded that this interlocutory order could not be reviewed by the Supreme Court. Id. at 148^19 (explaining that the remedy was “not one of those remedies over which” the Court was given “supervising power ... by law”).
Ordinarily, an inferior federal court is bound by the holdings of the Supreme Court. McCoy v. Mass. Inst. of Tech., 950 F.2d 13, 19 (1st Cir.1991). Here, however, it is doubtful that Pickett’s Heirs has any precedential force beyond its specific facts. The Court there explicitly admonished that it was not “called upon to decide” whether the case before it “was a case proper for the application of [the coram nobis] remedy.” 32 U.S. at 147. That very question lies at the heart of the instant matter, as the government contends that, in the circumstances of this case, AEDPA’s gatek-eeping provisions foreclose any resort to coram nobis. We therefore do not regard Pickett’s Heirs as contradicting the general proposition that coram nobis orders are appealable.
This conclusion is reinforced by modern appellate practice. A canvass of the deci-sional law reveals that, in the more recent cases, coram nobis orders routinely have been reviewed by federal appellate courts. See, e.g., United States v. Sawyer, 239 F.3d 31, 36 (1st Cir.2001); United States v. Keane, 852 F.2d 199, 200 (7th Cir.1988). The Supreme Court itself has asserted appellate jurisdiction in a case like this one, that is, a case questioning the trial court’s power to issue a writ of error co-ram nobis. See United States v. Morgan, 346 U.S. 502, 513, 74 S.Ct. 247, 98 L.Ed. 248 (1954). And, finally, the Federal Rules of Appellate Procedure contemplate that coram nobis orders may be appealed as a matter of course. See Fed. R.App. P. 4(a)(1)(C) (“An appeal from an order granting or denying an application for a writ of error coram nobis is an appeal in a civil case for purposes of Rule 4(a).”). That rule has the force of law. See Local Union No. 38 v. Custom Air Sys., 333 F.3d 345, 348 (2d Cir.2003).
History furnishes a simple means for reconciling whatever conflict a jaundiced eye might perceive between the curious language in Pickett’s Heirs and modern appellate practice. Coram nobis, in its present incarnation, bears little resemblance to the writ as it existed in the early nineteenth century. We briefly trace its evolution.
The writ of error coram nobis originated in sixteenth-century England as an instru*93ment used by trial courts to correct their own fact-based errors. See Carlisle v. United States, 517 U.S. 416, 428-29, 116 S.Ct. 1460, 134 L.Ed.2d 613 (1996); United States v. Mayer, 235 U.S. 55, 68, 35 S.Ct. 16, 59 L.Ed. 129 (1914); see also 7 LaFave et al., Criminal Procedure § 28.1(c), at 143 (3d ed.2007). For the most part, its use was confined to civil cases. See, e.g., 7 LaFave et al., supra § 28.1(c), at 143; Lester B. Orfield, The Writ of Error Co-ram Nobis in Civil Practice, 20 Ya. L.Rev. 423, 427 (1934). But, in time the tectonic plates shifted and the utility of the writ expanded. It eventually grew beyond the correction of mistakes of fact and became flexible enough to reach fundamental legal errors. See Morgan, 346 U.S. at 507-08, 74 S.Ct. 247; Mayer, 235 U.S. at 68, 35 S.Ct. 16; see also 3 Wright, King & Klein, Federal Practice & Procedure § 592, at 687 (3d ed.2004).
As the writ of error coram nobis morphed to encompass more than exclusively “factual” mistakes, it also began to gain a foothold in the criminal docket. See Sawyer, 239 F.3d at 37. This transmogrification occurred in a leisurely fashion; the writ was not readily deployed in state criminal cases until well after the Supreme Court decided Pickett’s Heirs. See Abraham L. Freedman, The Writ of Error Co-ram Nobis, 3 Temp. L.Q. 365, 373 (1929) (placing in the late nineteenth century “the first criminal case of consequence in which the writ was successfully employed”). The writ’s migration to criminal cases in the federal courts was even slower; as late as 1914, the Supreme Court pointedly left open the question of the writ’s availability in criminal cases. See Mayer, 235 U.S. at 69, 35 S.Ct. 16. Not until 1954 was the legitimacy of the writ’s use in criminal cases confirmed by the Court. See Morgan, 346 U.S. at 512, 74 S.Ct. 247.
We now have come full circle: in the modern federal system, use of the writ of error coram nobis is confined to criminal cases. See id. at 505 n. 4, 74 S.Ct. 247; see also Fed.R.Civ.P. 60(e) (abolishing the writ in civil cases). To complete this transformation, in criminal cases the writ has been seen more and more as a means to challenge law-based errors. See, e.g., Sawyer, 239 F.3d at 37-38; United States v. Tucor Int'l, Inc., 189 F.3d 834, 836 n. 1 (9th Cir.1999); United States v. Rankin, 1 F.Supp.2d 445, 453 (E.D.Pa.1998).
Viewed against this historical backdrop, whatever the Pickett’s Heirs Court meant when it stated that it lacked “supervising power” over a grant of coram nobis relief, that statement is now an anachronism. Nearly two hundred years of doctrinal development have sapped it of any contemporary vitality.
The petitioner has a fallback position. He asseverates that even if the grant or denial of a writ of error coram nobis is deemed reviewable, the government lacks statutory authority to appeal such an order. This asseveration has a certain superficial allure. After all, the government can mount an appeal in a criminal case only in accordance with some express statutory authorization. United States v. Sanges, 144 U.S. 310, 312, 12 S.Ct. 609, 36 L.Ed. 445 (1892); United States v. Watson, 386 F.3d 304, 307 (1st Cir.2004); United States v. Horn, 29 F.3d 754, 768 (1st Cir.1994); see 15B Wright, Miller & Cooper, Federal Practice & Procedure § 3919.1, at 598 (2d ed.1991).
Close perscrutation discloses that the petitioner’s asseveration rests on a flimsy foundation: the dubious proposition that a writ of error coram nobis is part of the warp and woof of a criminal proceeding, subject to this traditional limitation on government appeals. To support that proposition, he cherry-picks a snippet in *94which the Morgan Court characterized the writ of error coram nobis as a “step in the criminal case and not, like habeas corpus where relief is sought in a separate case and record, the beginning of a separate civil proceeding.” 846 U.S. at 505 n. 4, 74 S.Ct. 247. In our view, Morgan’s footnote 4 cannot bear the heavy weight that the petitioner piles upon it.
As an initial matter, the import of the footnote is tenebrous; the very next line of the footnote asserts that coram nobis “is of the same general character as [proceedings] under 28 U.S.C. § 2255.” Id. Section 2255 proceedings, like classic petitions for habeas corpus, are generally treated as civil in nature. See Heflin v. United States, 358 U.S. 415, 418 n. 7, 79 S.Ct. 451, 3 L.Ed.2d 407 (1959); Rogers v. United States, 180 F.3d 349, 352 n. 3 (1st Cir.1999).
In all events, the reach of Morgan’s footnote 4 is more circumscribed than the petitioner suggests. As Judge Friendly observed when facing a similarly ambitious construction of the Morgan footnote, the quoted language does not purport to dictate that coram nobis petitions must be treated as strictly criminal for purposes of appeal or otherwise. United States v. Keogh, 391 F.2d 138, 140 (2d Cir.1968). Rather, the Morgan Court was confronted with, and resolved, a much narrower problem: whether the writ could be deployed to correct errors in criminal cases notwithstanding the promulgation of Federal Rule of Civil Procedure 60(b) (which abolished coram nobis in “suits of a civil nature”). See Morgan, 346 U.S. at 505 n. 4, 74 S.Ct. 247. Judge Friendly found no reason to think that, in affirmatively answering this narrow question, the Morgan Court intended to determine that coram nobis orders should be treated as criminal for all purposes. Keogh, 391 F.2d at 140.
Like Judge Friendly, we decline to read the quoted snippet as extending beyond the question that the Morgan Court was endeavoring to answer. In taking this commonsense approach to Morgan’s footnote 4, we join several other courts that have followed Judge Friendly’s lead. See, e.g., United States v. Craig, 907 F.2d 653, 656-57 (7th Cir.1990); United States v. Cooper, 876 F.2d 1192, 1194 (5th Cir.1989); United States v. Balistrieri, 606 F.2d 216, 220-21 (7th Cir.1979); see also Fed. R.App. P. 4(a)(1)(C) & advisory committee notes.
Of course, this leaves an unanswered question: Should coram nobis be treated as civil or criminal for purposes of appeal? The answer to that question cannot be gleaned by grasping at convenient labels. While the modern writ of error coram nobis is used in criminal cases, its character necessarily reflects the vestiges of its origins in civil litigation. Accordingly, many courts have treated coram nobis as civil for a variety of purposes, such as the availability of discovery, see, e.g., Balistrieri, 606 F.2d at 221, and the temporal limits for filing notices of appeal, see, e.g., Keogh, 391 F.2d at 140. Given the lessons of history, we are convinced that coram nobis proceedings are best seen as hybrids — quasi-civil and quasi-criminal. See Craig, 907 F.2d at 656; Mercado v. United States, 183 F.2d 486, 487 (1st Cir.1950). On this view, the denomination of the nature of a given petition calls for a functional analysis rather than for doctrinal rigidity.
We see no reason here to apply the more restrictive rules governing criminal appeals to the government’s appeal in a coram nobis proceeding (even though that proceeding is ancillary to a criminal case). The traditional restrictions on the government’s right to appeal stem primarily from double jeopardy concerns. See *95Kepner v. United States, 195 U.S. 100, 133, 24 S.Ct. 797, 49 L.Ed. 114 (1904). Such concerns are far less weighty when dealing with collateral challenges to criminal convictions. See 15B Wright, Miller & Cooper, supra § 3919.9, at 732 (“There is less need to restrict government appeal opportunities in proceedings that come ... after the formal prosecution.”). Consequently, we hold that coram nobis proceedings are appealable as civil matters. Thus the government, like any other party, may appeal the grant or denial of the writ as a final order under 28 U.S.C. § 1291. See 15B Wright, Miller & Cooper, supra § 3919.9, at 740 (positing by analogy to habeas corpus that the government “should be allowed to appeal final disposition of coram nobis proceedings”); see also Fed. R.App. P. 4(a)(1)(C).
The petitioner has one last arrow in his jurisdictional quiver. He contends that even if we find appellate jurisdiction in theory, the government’s appeal here is time-barred. Once again, this contention derives from the Morgan Court’s cryptic characterization of coram nobis as a “step in the criminal case.” 346 U.S. at 505 n. 4, 74 S.Ct. 247. On that premise, the petitioner concludes that the government was subject to the 30-day filing deadline applicable to government appeals in criminal cases, Fed. R.App. P. 4(b)(1)(B); that the 30-day period ran from February 20, 2007 (when the district court announced its decision to grant the writ); and that, therefore, the government’s notice of appeal, filed in mid-April, was late.
This plaint is as insubstantial as a house built upon the shifting sands. The short answer to it is that the Federal Rules of Appellate Procedure explicitly provide that “[a]n appeal from an order granting or denying an application for a writ of error coram nobis is an appeal in a civil case” and, thus, is subject to a 60-day filing period. Fed. R.App. P. 4(a)(1)(C); see also Fed. R.App. P. 4(a)(1)(B). This language was added to the Appellate Rules in 2002 to resolve a conflict in the courts of appeals regarding the time limits that applied to appeals from coram nobis orders. See Fed. R.App. P. 4 advisory committee notes. The amended rule controls here. Accordingly, the government’s notice of appeal was timely even if, as the petitioner contends, the appeal period ran from February 20, 2007.5
B. AEDPA Preclusion.
Having confirmed the existence of appellate jurisdiction, we turn to the substance of the government’s appeal. Distilled to its essence, the argument here is that the petition was a wolf in sheep’s clothing: an unapproved second or successive section 2255 petition masquerading as a petition for a writ of error coram nobis. Relatedly, the government insists that coram nobis is unavailable to a prisoner in federal custody. If these arguments are well-founded, the district court lacked jurisdiction to entertain the petition.
The petitioner’s initial rejoinder is in the nature of a preemptive strike. He says that because the government failed to respond to his petition in a timeous fashion, *96it has waived or forfeited these arguments. This strike misfires for two reasons.
First, each of the government’s arguments goes to the district court’s subject-matter jurisdiction. This is important because parties cannot confer subject-matter jurisdiction on a district court by sloth or acquiescence. See United States v. Cotton, 535 U.S. 625, 630, 122 S.Ct. 1781, 152 L.Ed.2d 860 (2002); Espinal-Dominguez v. Puerto Rico, 352 F.3d 490, 495 (1st Cir.2003); Horn, 29 F.3d at 767-68.
Second, these arguments were raised, albeit belatedly, in the district court. That court chose not to treat them as waived but, rather, seemingly rejected the government’s motion for reconsideration on the merits (thus effectively rejecting the preclusion argument contained therein). The court made no comment about the timeliness vel non of the government’s proffer. Where a trial court chooses to overlook the belated nature of a filing and adjudicate the tardy claim or defense on the merits, that claim or defense may be deemed preserved for purposes of appellate review. See Negrón-Almeda v. Santiago, 528 F.3d 15, 26 (1st Cir.2008).
This brings us to the post-AEDPA framework for collateral review. The writ of habeas corpus historically has served as the principal vehicle for testing the legality of executive detentions. Attempting to address “practical problems that had arisen in the administration of the federal courts’ habeas corpus jurisdiction,” United States v. Hayman, 342 U.S. 205, 210, 72 S.Ct. 263, 96 L.Ed. 232 (1952), Congress enacted 28 U.S.C. § 2255 as a substitute for the traditional habeas remedy with respect to federal prisoners. See Act of June 25, 1948, 62 Stat. 967. This “virtual habeas” statute supplies a remedy that bears a strong family resemblance to the traditional writ of habeas corpus. The statute was intended to provide a federal prisoner with an exclusive means of challenging the validity of his conviction or sentence, save only in those few instances in which the statutory remedy proved “inadequate or ineffective to test the legality of his detention.” 28 U.S.C. § 2255(e). In this context, “inadequate or ineffective” requires an ex ante appraisal. Taylor v. Gilkey, 314 F.3d 832, 835-36 (7th Cir.2002) (collecting cases).
In 1996, Congress enacted AEDPA. That statute imposed significant new constraints on proceedings under section 2255.6 Some of these constraints were temporal; for example, AEDPA established a one-year statute of limitations for filing a section 2255 petition. 28 U.S.C. § 2255(f). Some of these constraints were numerical; for example, AEDPA required a federal prisoner who sought to prosecute a second or successive section 2255 petition to obtain pre-clearance, in the form of a certificate, from the court of appeals. Id. § 2255(h). By the terms of the statute, such a certificate will be made available only if the prisoner can show that the proposed second or successive petition is based either on newly discovered evidence or a new rule of constitutional law. Id. We have interpreted this provision as “stripping the district court of jurisdiction over a second or successive habeas petition unless and until the court of appeals has decreed that it may go forward.” Pratt v. United States, 129 F.3d 54, 57 (1st Cir.1997).
Although section 2255, as amended, provides a comprehensive remedial scheme *97for post-conviction relief, that scheme perpetuates the savings clause contained in section 2255(e). Moreover, the Supreme Court has made it pellucid that section 2255 does not preempt the entire array of common-law writs authorized under the All Writs Act, 28 U.S.C. § 1651. See Morgan, 346 U.S. at 510-11, 74 S.Ct. 247. Still, each attempted use of an extraordinary writ in connection with post-conviction relief must be judged on its own merits. The strictures of section 2255 cannot be sidestepped by the simple expedient of resorting to some more exotic writ. Put bluntly, the All Writs Act is not a jujube. At most, it constitutes “a residual source of authority to issue writs that are not otherwise covered by statute.” Pa. Bureau of Corr. v. U.S. Marshals Serv., 474 U.S. 34, 43, 106 S.Ct. 355, 88 L.Ed.2d 189 (1985).
We think it follows that when a statute — like section 2255 — specifically addresses a particular class of claims or issues, it is that statute, not the All Writs Act, that takes precedence. See id. The armamentarium of common-law writs, including the writ of error coram nobis, is thus available only to fill whatever interstices exist in the post-conviction remedial scheme made available to federal prisoners by way of section 2255. See United States v. Ayala, 894 F.2d 425, 428 (D.C.Cir.1990).
This gap-filling approach makes it essential for courts to plot, and then to patrol, the boundaries between section 2255 and the universe of old common-law writs. Otherwise, artful pleaders will tiptoe around those boundaries and frustrate Congress’s discernible intent. See, e.g., United States v. Winestock, 340 F.3d 200, 203 (4th Cir.2003); cf. Calderon v. Thompson, 523 U.S. 538, 553, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998) (patrolling boundaries of section 2254). In carrying out this sentry duty, courts must be guided by the precept that substance trumps form. See Rodwell v. Pepe, 324 F.3d 66, 71 (1st Cir.2003) (predicting that whether AEDPA’s limitations apply “will depend not on the label affixed to a particular motion but on its essence”). Thus, “[a]ny motion filed in the district court that imposed the sentence, and substantively within the scope of § 2255 ¶ 1, is a motion under § 2255, no matter what title the prisoner plasters on the cover.” Melton v. United States, 359 F.3d 855, 857 (7th Cir.2004) (emphasis in original).
Following this approach, courts regularly have recharacterized imaginatively captioned petitions to reflect that they derive their essence from section 2255 and, thus, must satisfy that section’s gatekeeping provisions. See, e.g., Winestock, 340 F.3d at 206-208 (recharacterizing a self-styled Rule 60(b) motion); United States v. Torres, 282 F.3d 1241, 1246 (10th Cir.2002) (recharacterizing a self-styled writ of co-ram nobis); Spivey v. State Bd. of Pardons and Paroles, 279 F.3d 1301, 1303 (11th Cir.2002) (recharacterizing a self-styled § 1983 action); Jiminian v. Nash, 245 F.3d 144, 149 (2d Cir.2001) (recharac-terizing a self-styled § 2241 petition); Ayala, 894 F.2d at 430 (recharacterizing a so-called writ of audita querela); United States v. Mosquera, 845 F.2d 1122, 1123 n. 1 (1st Cir.1988) (recharacterizing a self-styled Rule 35 motion).
A close analysis of the substance of the petition in this case leaves no doubt but that, regardless of its label, the petition falls within the compass of section 2255. On its face, the petition is brought on behalf of a federal prisoner still in custody and challenges his sentence as unauthorized under the statutes of conviction. This is a classic habeas corpus scenario, squarely within the heartland carved out by Congress in section 2255. See 28 U.S.C. § 2255(a) (offering a means of post-conviction relief to “[a] prisoner in custody *98under sentence of a [federal] court” who claims “that [his] sentence was in excess of the maximum authorized by law”). As such, the claim embodied in the petition is one cognizable in a section 2255 proceeding. Therefore, the district court should have recharacterized the petition as a section 2255 petition and proceeded accordingly.
The conclusion that the petition should have been treated as one arising under section 2255 sounds a death knell for the proceeding. The petition was filed more than ten years after the petitioner’s conviction had become final and more than five years after an earlier section 2255 petition (itself deemed to be time-barred). No pre-clearance had been either sought or obtained in this court. Consequently, the petition was an unauthorized second or successive habeas petition and was foreclosed on that basis. See 28 U.S.C. § 2255(h); see also United States v. Barrett, 178 F.3d 34, 42-45 (1st Cir.1999). Given this reality, the district court had only two choices: either dismiss the re-characterized petition or transfer it to this court. See Pratt, 129 F.3d at 57. Either way, the case was at an end: dismissal would have ended it, and transferring the recharacterized petition to this court would have been an exercise in futility. After all, the petition was not premised on either newly discovered evidence or a new rule of constitutional law. See 28 U.S.C. § 2255(h).
The petitioner attempts to escape the vise-like grip of AEDPA’s gatekeeping provisions by asserting that a writ of error coram nobis comprises an alternative, and constitutionally necessary, method for challenging an illegal sentence, at least when other means of collateral review have been exhausted. But where, as here, a petitioner’s claim falls within the heartland of section 2255, this assertion lacks force. “The writ of coram nobis may not be used to circumvent the clear congressional directive embodied in the ‘second or successive’ provisions of § 2255.” Barrett, 178 F.3d at 55.
If more were needed — and we doubt that it is — the writ of error coram nobis, in its modern form, is ordinarily available only to a criminal defendant who is no longer in custody. See United States v. Esogbue, 357 F.3d 532, 534 (5th Cir.2004); Obado v. New Jersey, 328 F.3d 716, 718 (3d Cir.2003); Torres, 282 F.3d at 1245; see also 3 Wright, King & Klein, supra § 592, at 689. The petitioner’s exercise in legal taxonomy therefore fails: the gap-filling allowed by the Morgan Court does not permit an inquiring court “to redefine a common law writ in order to create relief not otherwise available in the federal postconviction remedial scheme.” Ayala, 894 F.2d at 429 (emphasis in original); accord Doe v. INS, 120 F.3d 200, 203-04 (9th Cir.1997).
To be sure, we suggested in Barrett, 178 F.3d at 56, that there may be situations in which section 2255 does not completely occupy the field — but that suggestion was limited to those situations in which section 2255’s savings clause applies; that is, those situations in which the remedy provided by section 2255 is deemed “inadequate or ineffective.” Id. (quoting 28 U.S.C. § 2255(e)). This case does not come within the narrow confines of the savings clause.
As indicated above, section 2255 is both adequate and effective to address a claim of the type and kind mounted by the petitioner. Although the petitioner might now be unable to pursue that claim via a second or successive petition because of the combination of his own tardiness and AEDPA’s gatekeeping provisions, the mere inability to satisfy those require*99ments does not afford access to the savings clause. See id. at 50. To rule otherwise would reduce AEDPA’s gatekeeping provisions to “a meaningless gesture.” Id.; accord In re Davenport, 147 F.3d 605, 608 (7th Cir.1998).
The bottom line is that adequacy and effectiveness must be judged ex ante. Accordingly, post-conviction relief can be termed “inadequate” or “ineffective” only when, in a particular case, the configuration of section 2255 is such “as to deny a convicted defendant any opportunity for judicial rectification.” In re Davenport, 147 F.3d at 611 (emphasis in original).
The record here makes manifest that the petitioner cannot satisfy that demanding standard. He had multiple opportunities to challenge the imposition of his life sentences as inconsistent with the terms of the statutes of conviction. His alleged claim of error was available to him on both direct and collateral review;7 and those review modalities, if timely engaged, would have offered an adequate and effective means of rectifying the error. That he neglected to avail himself of these opportunities in no way detracts from their adequacy and effectiveness.
The fact that courts have allowed recourse to the savings clause in rare and exceptional circumstances, such as those in which strict adherence to AEDPA’s gatek-eeping provisions would result in a “complete miscarriage of justice,” In re Dorsainvil, 119 F.3d 245, 251 (3d Cir.1997), does not help the petitioner. Most courts have required a credible allegation of actual innocence to access the savings clause. See, e.g., Reyes-Requena v. United States, 243 F.3d 893, 904 (5th Cir.2001). The petitioner’s claim does not qualify; his petition does not posit that he is actually innocent of the crimes for which he was indicted, convicted, and sentenced.
Equally as important, the nature of the petitioner’s claim of error is incompatible with engagement of the savings clause. He does not charge that a life sentence is beyond the statutory maximum for the crimes of conviction. Rather, his claim is procedural: that the statutes of conviction required a jury recommendation in order to impose a life sentence, and that the judge, not the jury, made the recommendation in his case. This claim, if true, does not suggest a miscarriage of justice. We explain briefly.
The Supreme Court has defined the term “miscarriage of justice” as encompassing only those “extraordinary instances when a constitutional violation probably has caused the conviction of one innocent of the crime.” McCleskey v. Zant, 499 U.S. 467, 494, 111 S.Ct. 1454, 113 L.Ed.2d 517 (1991). The petitioner makes no such claim; he does not challenge the sufficiency of the evidence upon which the sentencing judge acted, nor does he suggest that it was likely that a jury would have reached a different result.
Rules that allocate decisionmaking responsibility between judge and jury are generally thought to be procedural. See Schriro v. Summerlin, 542 U.S. 348, 353, 124 S.Ct. 2519, 159 L.Ed.2d 442 (2004). As such, misalloeation of that responsibility is unlikely to implicate the accuracy or fundamental fairness of judicial proceedings. See, e.g., Sepulveda v. United States, 330 F.3d 55, 61 (1st Cir.2003) (holding, in the Apprendi context, that decisions *100made by the judge instead of the jury are “not the sort of error that necessarily undermines the fairness ... of judicial proceedings”) (internal quotation marks omitted). Here, the petitioner has given us no reason to suspect that the misallocation affected his substantial rights.
The final shot in the petitioner’s sling is that the right to correct an illegal sentence is a basic right for which the Constitution guarantees a remedy. This platitude has little utility here.
It is well-established that a criminal defendant possesses no general constitutional right to appellate review. See, e.g., McKane v. Durston, 153 U.S. 684, 687, 14 S.Ct. 913, 38 L.Ed. 867 (1894); United States v. Puzzanghera, 820 F.2d 25, 26-27 (1st Cir.1987). By the same token, errors committed at trial, even those of constitutional dimension, do not invariably require reversal. See, e.g., Washington v. Recuenco, 548 U.S. 212, 218, 126 S.Ct. 2546, 165 L.Ed.2d 466 (2006); see also Neder v. United States, 527 U.S. 1, 8, 119 S.Ct. 1827, 144 L.Ed.2d 35 (1999) (explaining that “most constitutional errors can be harmless”) (quoting Arizona v. Fulminante, 499 U.S. 279, 306, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991)).
Furthermore — as we already have explained — the petitioner had several opportunities for judicial review. If he failed to raise a particular claim of error at the appropriate time, enforcement of the usual rules of waiver and preclusion scarcely can be deemed a violation of his constitutional right to an appeal. Thus we conclude, without serious question, that to the extent that the Constitution affords a defendant a right to correct errors in judicial proceedings through either appellate review or collateral attack, that right has not been sullied here.
III. CONCLUSION
We appreciate the able district judge’s desire to correct an apparent error attributable to the lawyers’ shared misperception. We admire as well the ingenuity displayed by appointed counsel in attempting to rectify that mistake. But the rule of law must remain paramount and, as Lord Campbell warned long ago, hard cases have a propensity to make bad law. See Burnham v. Guardian Life Ins. Co., 873 F.2d 486, 487 (1st Cir.1989) (quoting United States v. Clark, 96 U.S. 37, 49, 24 L.Ed. 696 (1877) (Harlan, J., dissenting) (quoting Lord Campbell in East Indian Co. v. Paul, 7 Moo. P.C.C. 111)). That propensity must be held in check. That is our obligation here.
Implicit in the scheme created by AED-PA is the notion that certain claims, which might have been fruitful if timely asserted, may be foreclosed when a convicted defendant sleeps upon his rights. See Dodd v. United States, 545 U.S. 353, 359, 125 S.Ct. 2478, 162 L.Ed.2d 343 (2005); Jamison v. United States, 244 F.3d 44, 47 (1st Cir.2001). That our criminal justice system tolerates a certain risk of error might be of concern to some, but finality is indispensable to the proper functioning of that system. Under AEDPA, there is typically “only one bite at the post-conviction apple.” Barrett, 178 F.3d at 57. The petitioner has had that bite.
We need go no further. For the reasons elucidated above, the order granting the writ of error coram nobis is reversed, the writ is quashed, and the case is remanded to the district court with directions to vacate the amended judgment in the criminal case and reinstate the original sentence.
So Ordered.