On July 27, 1993, defendant Orlando Beals pleaded guilty to submitting false claims for federal income tax refunds in violation of 18 U.S.C. § 286. He was sentenced to 30 months' imprisonment followed by three years of supervised release. He finished his prison term on January 20, 1995 and began his supervised release. By September 1995, however, Beals had skipped many of his required drug testing sessions, and two of the times he did show up he tested positive for cocaine use. For these reasons and others, Judge Lindberg (1) revoked Beals' supervised release and ordered him to spend ten months in prison pursuant to 18 U.S.C. § 3583(e)(3); and (2) ordered that Beals continue on supervised release following imprisonment pursuant to 18 U.S.C. § 3583(h). Beals asserts that because Section 3583(h) was enacted subsequent to his conviction, ordering him to undergo a term of supervised release after his ten-month prison term violates the Ex Post Facto Clause of the Constitution (Article I, § 9, cl. 3). We agree.

The statute regarding revocation of supervised release in effect at the time of Beals' conviction was 18 U.S.C. § 3583(e). That section stated that when a defendant violated the terms of his supervised release, a district court was allowed to:

(1) terminate the term of supervised release ...;

(2) extend the term of supervised release if less than the maximum term was previously imposed, and ... modify, reduce or enlarge the conditions of supervised release ...;

(3) revoke the term of supervised release, and require the person to serve in prison all or part of the term of supervised release, without credit for time previously served on postrelease supervision ...; or

(4) order the person to remain at his place of residence.

In United States v. McGee, 981 F.2d 271, 274 (7th Cir.1992), we joined the majority of Circuit Courts of Appeals in holding that Section 3583(e)(3) did not allow the district court that revoked a defendant's supervised release to impose an additional term of supervised release following a period of incarceration. Most likely in response to McGee and cases that reached a similar conclusion, Congress enacted 18 U.S.C. § 3583(h) on September 13, 1994. That subsection now provides:

When a term of supervised release is revoked and the defendant is required to serve a term of imprisonment that is less than the maximum term of imprisonment authorized under subsection (e)(3), the court may include a requirement that the defendant be placed on a term of supervised release after imprisonment. The length of such a term of supervised release shall not exceed the term of supervised release authorized by statute for the offense that resulted in the original term of supervised release, less any term of imprisonment that was imposed upon revocation of supervised release.

Pursuant to this new subsection, Judge Lindberg ordered that Beals spend two years in supervised release following his ten-month prison term (a period within the subsection's parameters). The question we must answer is whether that sentence was constitutional in light of the fact that Beals' original offense took place prior to that subsection's enactment.

Initially, we address the government's contention that it is impossible to apply Subsection (h) retroactively because its enactment was not a change in the law, but merely a "clarification" of it.1 In other words, the government argues that Subsection (h) represents the "true" state of the law regarding supervised release revocation prior to September 13, 1994, despite the fact that this Circuit (and eight others) had not interpreted Section 3583(e) to allow what Subsection (h) now expressly permits. This assertion is in direct conflict with Rivers v. Roadway Express, Inc., 511 U.S. 298, ----, 114 S. Ct. 1510, 1516, 128 L. Ed. 2d 274, where the Supreme Court specifically rejected the contention that amending a statute merely "corrects" the judicial construction of the law. "A change in statutory language--or ... a new statutory section--does not imply that the exegesis of the prior law was mistaken.... Congress acts by legislating rather than by reinterpreting laws already on the books." Evans v. United States Parole Commission, 78 F.3d 262, 266 (7th Cir.1996). Section 3583(h) represented a change in the law of this Circuit, and we will properly address it as such.

Article I of the Constitution provides that neither Congress nor any State shall pass any "ex post facto Law." See Art. I, § 9, cl. 3. Our understanding of what is meant by "ex post facto" largely derives from Calder v. Bull, 3 Dall. 386, 1 L. Ed. 648 (1798), in which the Court summarized which laws fell "within the words and intent of the prohibition":

1st. Every law that makes an action done before the passing of the law, and which was innocent when done, criminal; and punishes such action. 2d. Every law that aggravates a crime, or makes it greater than it was, when committed. 3d. Every law that changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed. 4th. Every law that alters the legal rules of evidence, and receives less, or different testimony, than the law required at the time of the commission of the offense, in order to convict the offender. Id. at 390.

Beals argues that Subsection (h) falls within the third of the above prohibitions, inflicting a greater punishment than the law annexed to his crime at the time he committed it. Since Calder, a two-pronged test has been developed to determine whether the application of a law violates the Ex Post Facto Clause: (1) The law must apply to events occurring before its enactment; and (2) the law must disadvantage the offender affected by it. Miller v. Florida, 482 U.S. 423, 430, 107 S. Ct. 2446, 2451, 96 L. Ed. 2d 351.

Taking the second element first, we conclude that Subsection (h) could potentially disadvantage Beals. Prior to its enactment, if a defendant violated the terms of his supervised release, a district court was empowered to revoke the supervised release and send the defendant back to prison for a period equal to all or part of the maximum term of supervised release authorized for his original offense. 18 U.S.C. § 3583(e)(3). After Subsection (h), a district court can still revoke the supervised release and send the defendant to prison for the same amount of time. However, if the district court sends the defendant back to prison for a period of time less than the maximum allowed, the court can additionally require the defendant to spend the balance of the maximum time on another term of supervised release:

Under [Subsection (h) ], the maximum length of additional supervised release term may not exceed the maximum period of supervised release authorized by statute for the offense, less any term of imprisonment imposed upon revocation. For example, in the case of a Class C felony for which the maximum supervised release term is three years, a defendant who is revoked and re-imprisoned for 18 months could be ordered to serve as much as 18 additional months on supervised release (36-month maximum term of supervised release minus 18 months imprisonment equals 18 months possible re-release supervision). 137 Cong. Rec. S7769-01, S7771.

At first glance, it is difficult to see how Subsection (h) operates as a disadvantage: Prior to its enactment the defendant could go to prison for up to the maximum time period allowed; after its enactment the defendant can undergo a combination of prison and supervised release for up to the maximum time period allowed. Spending time on supervised release is hardly worse to the defendant than spending time in prison. The disadvantage comes from the fact that a defendant receives no credit toward the imprisonment cap for "time previously served on postrelease supervision." 18 U.S.C. § 3583(e)(3). Prior to Subsection (h), a defendant could serve only one term of supervised release, and thus only once "lose" credit for time served prior to the revocation. After Subsection (h), a defendant can serve multiple terms of supervised release and thus potentially "lose" multiple periods of time after the initial revocation:

If the same defendant [as in the example above] was again revoked, he could be re-imprisoned for not exceeding six months (24-month cap minus 18 months previously-served imprisonment equals 6 months allowable imprisonment) and if so imprisoned, could not thereafter be placed on supervision (because the two-year imprisonment cap would have been reached). Thus, under [Subsection (h) ], a defendant would always be credited for incarceration time against both the cap on re-imprisonment and the maximum authorized period of supervised release. 137 Cong. Rec. S7769-01, S7771.

The following hypothetical illustrates the point. Assume that Defendant A is convicted of a Class C felony and sentenced to a term of imprisonment followed by three years of supervised release (the maximum allowed under 18 U.S.C. § 3583(b)(2)). He serves his prison time and is released under supervision. One year into his supervised release period, he violates the terms of the release. Prior to Subsection (h), because an additional term of supervised release was not permitted, the maximum penalty the court could impose was two years imprisonment. 18 U.S.C. § 3583(b)(3). At the end of two years the government's supervision of A is extinguished. After Subsection (h), the district court, perhaps believing itself more lenient, may order A to serve two years on a combination of imprisonment and supervised release, say one year in prison and one year on supervised release. If A then violates the terms of that second supervised release six months into it, the court has the power to send him back to prison again, this time for up to one year (the two-year maximum minus the one-year term of imprisonment he has already served). Under this scenario, A's punishment totals two and a half years from the time of his initial revocation (one year in prison, six months on supervised release, and then another year in prison)--six months longer than that allowed prior to Subsection (h). And the potential exists for even greater discrepancies.

Having concluded that Subsection (h) can act to Beals' disadvantage, we now address the question of retroactive application. A law is retroactive if it "changes the legal consequences of acts completed before its effective date." Weaver v. Graham, 450 U.S. 24, 31, 101 S. Ct. 960, 965-66, 67 L. Ed. 2d 17. We thus must decide whether the punishment imposed upon Beals' revocation "should be considered the continuing 'legal consequence' of [Beals'] original crimes, or viewed instead as the independent 'legal consequence' of [Beals' later] misconduct." United States v. Reese, 71 F.3d 582, 584 (6th Cir.1995).

The difficulty in answering the above question arises from two lines of cases involving essentially the same scenario as that before us: (1) a defendant commits an offense ("original offense"); (2) the law is changed such that the original offense allows a new or increased punishment if the defendant subsequently engages in certain conduct ("subsequent conduct"); and (3) the defendant then engages in the subsequent conduct and is assessed the new or increased punishment. See Reese, 71 F.3d at 588. The first line of cases involves recidivist statutes, which enhance the subsequent sentences of repeat offenders. In doing so, the statutes often count as predicate crimes those offenses committed before the statutes' effective date. Courts have consistently viewed these situations not as increased punishment for the original offense, but as punishment only for the later conduct. For example, in Gryger v. Burke, 334 U.S. 728, 68 S. Ct. 1256, 92 L. Ed. 1683, the Supreme Court upheld against ex post facto challenge a life sentence for a defendant who had been convicted four times, where the first of the defendant's convictions predated the enactment of the recidivist law:

The sentence as a fourth offender or habitual criminal is not to be viewed as ... additional penalty for the earlier crimes. It is a stiffened penalty for the latest crime, which is considered to be an aggravated offense because a repetitive one. Gryger, 334 U.S. at 732, 68 S.Ct. at 1258-59.

The second line of cases involves parole violations. The seminal case in that line is Greenfield v. Scafati, 277 F. Supp. 644 (D.Mass.1967) (three-judge court), aff'd. mem., 390 U.S. 713, 88 S. Ct. 1409, 20 L. Ed. 2d 250. In Greenfield, the defendant was sentenced to 5-7 years in prison for his original crime. At the time of sentencing, Massachusetts law allowed for the accumulation of "good-conduct" credits while in prison, by which a defendant was able to advance the date of his release from prison. While Greenfield was serving his sentence, Massachusetts amended the relevant statute to provide for the forfeiture of good-conduct credits for parole violations. Greenfield was eventually paroled, violated that parole, and forfeited the good-conduct credits he had accumulated. The court concluded that by changing the consequences of a parole violation, Massachusetts had increased the punishment of the defendant's original sentence and thus violated the Ex Post Facto Clause of the Constitution. Greenfield, 277 F. Supp. at 646. The Supreme Court affirmed the decision without issuing an opinion. Greenfield, 390 U.S. 713, 88 S. Ct. 1409, 20 L. Ed. 2d 250. Several courts have subsequently followed Greenfield and held that statutes forfeiting good-time credits for parole violations cannot be applied to those defendants whose underlying offenses took place before the statutes' enactment. See, e.g., Williams v. Lee, 33 F.3d 1010 (8th Cir.1994), certiorari denied, --- U.S. ----, 115 S. Ct. 1393, 131 L. Ed. 2d 244; Fender v. Thompson, 883 F.2d 303 (4th Cir.1989); Beebe v. Phelps, 650 F.2d 774, 775-776 (5th Cir. Unit A July 1981) (per curiam); Shepard v. Taylor, 556 F.2d 648 (2d Cir.1977).

The similarity between punishment for supervised release violations, parole violations, and recidivist conduct has created a split in the Circuits. The Second, Fourth, and Ninth Circuits found the punishment for supervised release violations similar to that imposed for parole violations. Thus those Circuits followed Greenfield and concluded that it was an ex post facto violation to apply 18 U.S.C. § 3583(g)2 to a defendant whose original offense occurred prior to that Section's enactment. United States v. Meeks, 25 F.3d 1117 (2d Cir.1994); United States v. Paskow, 11 F.3d 873 (9th Cir.1993); United States v. Parriett, 974 F.2d 523 (4th Cir.1992). The Sixth Circuit disagreed, determining that the punishment was similar to that imposed under a recidivist statute: "[the defendant] is merely being required to serve time not for the original crime, but for the violation of supervised release." Reese, 71 F.3d at 590. Thus that Circuit followed the Supreme Court's Gryger case and concluded that it was not an ex post facto violation to apply Section 3583(g) to a defendant whose original offense took place before that Section was enacted.

At first glance, the parole violation cases and the recidivist statute cases appear to conflict: They reach opposite ex post facto conclusions despite the fact that the increased punishment for the subsequent conduct stems directly from the defendant's original offense. However, the cases are easily harmonized by considering why the government punishes the subsequent conduct in the first place. The increased punishment imposed under a recidivist statute is triggered by subsequent conduct that is itself a crime. The government punishes that conduct because of its nature, not because of the defendant's original offense. Therefore, it is logical to link the increased punishment only to the subsequent conduct for ex post facto purposes. For this reason, the recidivist cases correctly hold that recidivist statutes do not violate the Ex Post Facto Clause, despite taking into account pre-statute offenses in determining punishment. See Gryger, supra. With parole violations, however, the conduct resulting in punishment is often not a crime. Thus the government punishes the subsequent conduct only because of the defendant's original offense. Therefore, it is impossible to separate the punishment imposed for the subsequent conduct from the original offense and it makes sense to link the punishment to that offense for ex post facto purposes. Thus the parole cases correctly hold that statutes enacted after the original offense that increase the total punishment imposed violate the Ex Post Facto Clause. See Greenfield, supra.

Approaching supervised release violations with the above considerations in mind, the Sixth Circuit's position in Reese is untenable. Conduct that violates the terms of supervised release, like that of parole violations, is often not criminal. See, e.g., 18 U.S.C. §§ 3583(d), 3563(b)(1), (5), and (10) (defendant violates supervised release if he does not support his dependents, work conscientiously at suitable employment, and undergo medical treatment). Therefore, the government punishes that conduct only because of the defendant's original offense. For that reason, we must link the punishment imposed for the subsequent conduct to the original offense for ex post facto purposes.

We agree with the Sixth Circuit that supervised release and parole are technically different species of punishment: "Unlike parole, a term of supervised release does not replace a portion of the [original] sentence of imprisonment, but rather is an order of supervision in addition to any term of imprisonment imposed by the court." U.S.S.G. ch. 7, pt. A2(b). As a result of this technical difference, the relation between sentencing guidelines and the punishment imposed for a violation is different in parole situations than in supervised release situations: Whereas the term of imprisonment for parole violations is limited to the term allowable under the original offense, revocation of supervised release can result in cumulative punishment that greatly exceeds the maximum original prison sentence. See, e.g., United States v. Robinson, 62 F.3d 1282 (10th Cir.1995). However, this distinction is meaningless for purposes of ex post facto analysis. Under both systems, a defendant is sentenced for an original offense to a combination of imprisonment and post-imprisonment release. Any law enacted after the original offense that increases the total amount of time he can spend in that combination violates the Ex Post Facto Clause.

Applying the above principles to Beals' situation, it is clear that there was an ex post facto violation. Beals was originally convicted for submitting false claims for tax refunds and received a sentence of thirty months' imprisonment and three years' supervised release. At the time of his offense, because 18 U.S.C. § 3583(e) did not permit a district court to sentence a defendant to subsequent supervised release terms, the length of Beals' punishment was limited to his original term of imprisonment and the period of his supervised release (minus any time served on post-release prior to the revocation). As discussed above, Subsection (h) potentially increases the length of that punishment without Beals committing another crime. Thus it disadvantages him and its application violates the Ex Post Facto Clause.

Beals also claims that his ten-month term of imprisonment did not meet our plainly reasonable sentencing standard. See McGee, 60 F.3d at 1272. However, the record demonstrates that the district court correctly determined that Beals had a criminal history category of II. Given that his supervised release violations were of Grade C type, U.S.S.G. § 7B1.1(a)(3), the ten-month imprisonment term was within the range provided by the guidelines. U.S.S.G. § 7B1.4(a). The sentence was not unreasonable.

For the reasons discussed above, the case is remanded to the district court for it to amend its revocation order by eliminating the requirement that Beals serve a second term of supervised release following his term of imprisonment.