with whom The Chief Justice joins, dissenting.
Historians have traced the origins of our constitutional guarantee against double jeopardy, back to the days of Demosthenes, who stated that “the laws forbid the same man to be tried twice on the same issue. . . .” 1 Demosthenes 589 (J. Vince trans., 4th ed. 1970). Despite its roots in antiquity, however, this guarantee seems both one of the least understood and, in recent years, one . of the most frequently litigated provisions of the Bill of Rights. This Court has done. little to alleviate the confusion, and our opinions, including ones authored by me, are replete with mea cwlpo’s. occasioned by shifts in assumptions and emphasis. Compare, e. g., United States v. Jenkins, 420 U. S. 358 (1975), with United States v. Scott, 437 U. S. 82 (1978) (overruling Jenkins). See also Burks v. United States, 437 U. S. 1, 9 (1978) (Our *700holdings on this subject “can hardly be characterized as models of consistency and clarity”). Although today’s decision takes a tentative step toward recognizing what I believe to be the proper role for this Court in determining the permissibility of multiple punishments, it ultimately compounds the confusion that has plagued us in the double jeopardy area.
I
In recent years we have stated in the manner of “black letter law” that the Double Jeopardy Clause serves three primary purposes. First, it protects against a second prosecution for the same offense after an acquittal. Second, it protects against a second prosecution for the same offense after a conviction. Third, it protects against multiple punishments for the same offense. See North Carolina v. Pearce, 395 U. S. 711, 717 (1969); Brown v. Ohio, 432 U. S. 161, 165 (1977). See also ante, at 688 (opinion of the Court). Obviously, the scope of each of these three protections turns upon the meaning of the words “same offense,” a phrase deceptively simple in appearance but virtually kaleidoscopic in application. Indeed, we have indicated on at least one prior occasion that the meaning of this phrase may vary from context to context, so that two charges considered the same offense so as to preclude prosecution on one charge after an acquittal or conviction on the other need not be considered the same offense so as to bar separate punishments for each charge at a single proceeding. See Brown v. Ohio, supra, at 166-167, n. 6.
In the present case we are asked to decide whether the Double Jeopardy Clause bars the imposition of separate punishments for the crimes of rape and felony murder based on rape. Because the sentences challenged by petitioner were imposed at a single criminal proceeding, this case obviously is not controlled by precedents developed in the context of successive prosecutions. Thus, the Court rightly *701eschews reliance upon Harris v. Oklahoma, 433 U. S. 682 (1977), where we concluded that the crimes of robbery and felony murder predicated on that robbery were similar enough to prevent the State of Oklahoma from prosecuting a person for the former offense after convicting him of the latter offense. See ante, at 694 (opinion of the Court). See also ante, at 698-699 (Blackmun, J.,' concurring in judgment).
Having determined that this case turns on the permissibility of “multiple punishments” imposed at a single criminal proceeding, the Court takes a tentative step in what I believe to be the right direction by indicating that the “dis-positive question” here is whether Congress intended to authorize separate punishments for the two crimes. Ante, at 689 (opinion of the Court). As Mr. Justice Blackmun notes in his concurrence, this Court has not always been so forthright in recognizing that Congress could, if it so desired, authorize cumulative punishments for violation of two separate statutes, whether or not those statutes defined “separate offenses” in some abstract sense. See ante, at 698. While we have hinted at this proposition in prior opinions, see, e. g., Brown v. Ohio, supra, at 165; Gore v. United States, 357 U. S. 386, 394 (1958) (Warren, C. J., dissenting), we have just as often hedged our bets with veiled hints that a legislature might offend the Double Jeopardy Clause by authorizing too many separate punishments for any single “act.” See, e. g., Simpson v. United States, 435 U. S. 6, 11-12 (1978); Sanabria v. United States, 437 U. S. 54, 69 (1978); Jeffers v. United States, 432 U. S. 137, 155 (1977) (plurality opinion). To the extent that this latter thesis assumes that any particular criminal transaction is made up of a determinable number of constitutional atoms that the legislature cannot further subdivide into separate offenses, “it demands more of the Double Jeopardy Clause than it is capable of supplying.” Westen & Drubel, Toward a General Theory of Double Jeopardy, 1978 S. Ct. Rev. 81, 113. See also Note, Twice in Jeopardy, 75 Yale L. J. 262, 311-313 (1965).
*702Having come thus far with the Court and the concurrence, I here part company, for it seems clear to me that, if the only question confronting this Court is whether Congress intended to authorize cumulative punishments for rape and for felony murder based upon rape, this Court need decide no constitutional question whatsoever. Axiomatically, we are obligated to avoid constitutional rulings where a statutory ruling would suffice. See Hagans v. Lavine, 415 U. S. 528, 549 (1974); Ashwander v. TV A, 297 U. S. 288, 347 (1936) (Brandeis, J., concurring). Thus, to the extent that the trial court exceeded its legislative authorization in sentencing petitioner to consecutive sentences for rape and felony murder where Congress intended the offenses to merge, our holding should rest solely on our interpretation of the relevant statutes rather than on vague references to “the principle of the separation of governmental powers embodied in the Double Jeopardy Clause of the Fifth Amendment. . . Ante, at 695 (opinion of the Court).
Like many of the false trails we have followed in this area, the Court’s confusion of statutory and constitutional inquiries is not without precedent. Brown v. Ohio contains dictum to the effect that, “[w]here consecutive sentences are imposed at a single criminal trial,” the Double Jeopardy Clause prevents the sentencing court from “exceed[ing] its legislative authorization by imposing multiple punishments for the same offense.” 432 U. S., at 165. In support of this dictum, which I believe ill-considered, Brown cited three cases: Ex parte Lange, 18 Wall. 163 (1874); Bell v. United States, 349 U. S. 81 (1955); and Gore v. United States, supra. In doing so, it tied together three separate strands of cases in what may prove to be a true Gordian knot.
In Ex parte Lange petitioner had been convicted under a statute authorizing a punishment of either fine or imprisonment. The District Court nevertheless sentenced him to a fine and imprisonment. Petitioner had paid his fine and had begun to serve his sentence when the District Court, appar*703ently recognizing its mistake, held a new sentencing proceeding and resentenced him to imprisonment only. Noting that petitioner had fully satisfied the relevant statute by paying the fine, this Court held that he was entitled to protection from a second punishment “in the same court, on the same facts, for the same statutory offence.” 18 Wall., at 168. As is borne out by subsequent cases, the Double Jeopardy Clause as interpreted in Ex parte Lange prevents a sentencing court from increasing a defendant’s sentence for any particular statutory offense, even though the second sentence is within the limits set by the legislature. See North Carolina v. Pearce, 395 U. S. 711 (1969); United States v. Benz, 282 U. S. 304, 307 (1931). See also United States v. Sacco, 367 F. 2d 368 (CA2 1966); United States v. Adams, 362 F. 2d 210 (CA6 1966); Kennedy v. United States, 330 F. 2d 26 (CA9 1964).
In Bell v. United States, supra, this Court considered a question wholly different from that considered in Ex parte Lange and its progeny: the proper units into which a statutory offense was to be divided. The petitioner in Bell had been convicted of two counts of violating the Mann Act, 18 U. S. C. § 2421 et seq., for carrying two women across state lines for an immoral purpose. Both counts dealt with the same trip in the same car. The question presented to the Court was whether simultaneous transportation of more than one woman in violation of the Mann Act constituted multiple violations of that Act subjecting the offender to multiple punishments. The Court noted that Congress could, if it so desired, hinge the severity of the punishment on the number of women involved. Finding no evidence of such an intent, the Court applied the traditional “rule of lenity” and held that petitioner could only be punished for a single count.
Most significantly for our purposes, Bell was based entirely upon this Court’s interpretation of the statute and the relevant legislative intent; it did not mention the Double Jeopardy Clause at all. In finding congressional intent on the *704appropriate unit of prosecution dispositive, the Court acted consistently with a long line of cases based in English common law. In Crepps v. Durden, 2 Cowp. 640, 98 Eng. Hep. 1283 (K. B. 1777), Lord Mansfield, writing for a unanimous court, held that the sale of four loaves of bread on Sunday in violation of a statute forbidding such sale constituted one offense, not four. According to Lord Mansfield: “If the Act of Parliament gives authority to levy but one penalty, there is an end of the question. . . Id., at 646, 98 Eng. Rep., at 1287. One hundred years later, this Court expressly adopted the reasoning of Crepps that the proper unit of prosecution was completely dependent upon the legislature’s intent. See In re Snow, 120 U. S. 274, 283-286 (1887). We have consistently abided by this rule since that time, noting on at least one occasion that “[t]here is no constitutional issue presented” in such cases. See Ladner v. United States. 358 U. S. 169, 173 (1958). See also United States v. Universal C. I. T. Credit Corp., 344 U. S. 218 (1952); Ebeling v. Morgan, 237 U. S. 625 (1915). Cf. Sanabria v. United States, 437 U. S., at 69-70 (successive prosecutions).
Gore v. United States, the third case cited in Brown, presented an issue analogous to, but slightly different from, that presented in Bell and the other unit-of-proseeution cases, namely, the permissibility of consecutive sentences when a defendant committed a single act that violated two or more criminal provisions. This issue, the precise one confronting us today, has been litigated in an astonishing number of statutory contexts with little apparent analytical consistency. See, e. g., Simpson v. United States, 435 U. S. 6 (1978); Harris v. United States, 359 U. S. 19 (1959); Heflin v. United States, 358 U. S. 415 (1959); Prince v. United States, 352 U. S. 322 (1957); Pereira v. United States, 347 U. S. 1 (1954); American Tobacco Co. v. United States, 328 U. S. 781 (1946) ; Holiday v. Johnston, 313 U. S. 342 (1941); Blockburger v. United States, 284 U. S. 299 (1932); Morgan v. Devine, 237 U. S. 632 (1915); Burton v. United States, 202 U. S. 344 *705(1906); Carter v. MeClaughry, 183 U. S. 365 (1902). In some of these cases the Court seems to have recognized that it was attempting to divine legislative intent. See, e. g., Prince v. United States, supra, at 328; Morgan v. Devine, supra, at 638-639; Burton v. United States, supra, at 377. In other cases, the Court seemed to apply a “same evidence” test borrowed from cases involving successive prosecutions.1 See, e. g., Pereira v. United States, supra, at 9; Carter v. McClaughry, supra, at 394-395. In still others it is difficult to determine the precise basis for the Court’s decision. See, e. g., Harris v. United States, supra. As in the unit-of-prosecution cases, this Court has specified on at least one occasion that the erroneous imposition of cumulative sentences in a single case raises no constitutional issue at all. See Holiday v. Johnston, supra, at 349.
Unlike the Court, I believe that the Double Jeopardy Clause should play no role whatsoever in deciding whether cumulative punishments may be imposed under different statutes at a single criminal proceeding. I would analogize the *706present case to our unit-of-prosecution decisions and ask only whether Congress intended to allow a court to impose consecutive sentences on a person in petitioner’s position. To paraphrase Lord Mansfield’s statement in Crepps v. Durden, supra, that should be the end of the question. As even the Court’s analysis of the merits here makes clear, see ante, at 690-694, traditional statutory interpretation as informed by the rule of lenity completely supplants any possible additional protection afforded petitioner by the Double Jeopardy Clause.
The difference in this context between a constitutional decision and a statutory decision is not merely one of judicial semantics. Both the Court and the concurrence appear to invoke the Double Jeopardy Clause to justify their refusal to defer to the District of Columbia Court of Appeals’ interpretation of these locally applicable statutes. See ante, at 688 (opinion of the Court); ante, at 696 (Blackmun, J., concurring in judgment). The mischief in this approach, I believe, is well illustrated in a footnote — fairly described as either cryptic or tautological — stating that “[t]he Due Process Clause of the Fourteenth Amendment . . . would presumably prohibit state courts from depriving persons of liberty or property as punishment for criminal conduct except to the extent authorized by state law.” Ante, at 690, n. 4 (opinion of the Court). The effect of this and similar statements in the opinion of the Court, I fear, will be to raise doubts about questions of state law that heretofore had been thought to be exclusively the province of the highest courts of the individual States. To the extent that the Court implies that a state court can ever err in the interpretation of its own law and that such an error would create a federal question reviewable by this Court, I believe it clearly wrong.2 For the question in *707such cases is not whether the lower court “misread” the relevant statutes or its own common law, but rather who does the reading in the first place.
II
Because the question before us is purely one of statutory interpretation, I believe that we should adhere to our “longstanding practice of not overruling the courts of the District on local law matters ‘save in exceptional situations where egregious error has been committed.’ ” Pernell v. Southall Realty, 416 U. S. 363, 369 (1974), quoting from Griffin v. United States, 336 U. S. 704, 718 (1949). In the present case I would suggest that the lower court, far from committing “egregious error,” engaged in analysis much more sophisticated than that employed by the Court herein and reached a conclusion that is not only defensible, but quite probably correct.
The Court’s attempt to determine whether Congress intended multiple punishment in a case like petitioner’s is really quite cramped. - It looks first to the legislative history surrounding the adoption of the relevant provisions and finds that history inconclusive. See ante, at 690, and n. 5. It then attempts to mechanisticaEy apply the rule of statutory construction employed by this Court in Blockburger v. United States, 284 U. S. 299 (1932). See ante, at 691-694. Under that test, two statutory provisions are deemed to constitute the “same offense” so as to preclude imposition of multiple punishments unless “each provision requires proof of a fact which the other does not.” 284 U. S., at 304. In Blockburger, for example, this Court determined that a provision forbidding the sale of certain drugs except in or from the original stamped package and a provision forbidding the selling of the same drugs “not in pursuance of a written order of the” purchaser defined separate offenses because “each of the offenses created requires proof of a different element.” Ibid. Thus, separate penalties could be imposed under each statute, even though both offenses were based on the same sale.
*708Two observations about the Blockburger test are especially relevant in this case. First, the test is a rule of statutory construction, not a constitutional talisman.3 See Iannelli v. United States, 420 U. S. 770, 785, n. 17 (1975). Having already posited that the Double Jeopardy Clause imposes no restraint upon a legislature’s ability to provide for multiple punishments, I believe it clear that a legislature could, if it so desired, provide for separate punishments under two statutory provisions, even though those provisions define the “same offense” within the meaning of Blockburger. To take a simple example, a legislature might set the penalty for assault at two years’ imprisonment while setting the penalty for assault with a deadly weapon as “two years for assault and an additional two years for assault with a deadly weapon.” Even though the former crime is obviously a lesser included offense of the latter crime — or, in the rubric of Blockburger, the first offense does not require proof of any fact that the second does not — neither Blockburger nor the Double Jeopardy Clause would preclude the imposition of the “cumulative” sentence of two years.4
Second, the Blockburger test, although useful in identifying statutes that define greater and lesser included offenses in the traditional sense, is less satisfactory, and perhaps even misdirected, when applied to statutes defining “compound” and “predicate” offenses. Strictly speaking, two crimes do not stand in the relationship of greater and lesser included offenses unless proof of the greater necessarily entails proof of the *709lesser. See Brown v. Ohio, 432 U. S., at 167-168. See also Black’s Law Dictionary 1048 (rev. 4th ed. 1968). In the case of assault and assault with a deadly weapon, proof of the latter offense will always entail proof of the former offense, and this relationship holds true regardless whether one examines the offenses in the abstract or in the context of a particular criminal transaction.
On the other hand, two statutes stand in the relationship of compound and predicate offenses when one statute incorporates several other offenses by reference and compounds those offenses if a certain additional element is present. To cite one example, 18 U. S. C. § 924 (c)(1) states that “[w]hoever . . . uses a firearm to commit any felony for which he may be prosecuted in a court of the United States . . . shall ... be sentenced to a term of imprisonment for not less than one year nor more than ten years.” • Clearly, any one of a plethora of felonies could serve as the predicate for a violation of § 924 (c)(1).
This multiplicity of predicates creates problems When one attempts to apply Blockburger. If one applies the test in the abstract by looking solely to the wording of § 924 (c)(1) and the statutes defining the various predicate felonies, Block-burger would always permit imposition of cumulative sentences, since no particular felony is ever “necessarily included” within a violation of § 924 (c)(1). If, on the other hand, one looks to the facts alleged in a particular indictment brought under § 924 (c)(1), then Blockburger would bar cumulative punishments for violating § 924 (c)(1) and the particular predicate offense charged in the indictment, since proof of the former would necessarily entail proof of the latter.
Fortunately, in the case of § 924 (c)(1) Congress made its intention explicit, stating unequivocally that the punishment for violation of that statute should be imposed “in addition to the punishment provided for the commission of [the predicate] felony. . . .” 18 U. S. C. § 924 (c). But in the present *710case, where the statutes at issue also stand in the relationship of compound and predicate offenses, Congress has not stated its intentions so explicitly. The felony-murder statute under consideration here provides:
“Whoever, being of sound memory and discretion, kills another purposely, either of deliberate and premeditated malice or by means of poison, or in perpetrating or attempting to perpetrate any offense punishable by imprisonment in the penitentiary, or without purpose so to do kills another in perpetrating or in attempting to perpetrate any arson, . . . rape, mayhem, robbery, or kidnapping, or in perpetrating or attempting to perpetrate any housebreaking while armed with or using a dangerous weapon, is guilty of murder in the first degree.” D. C. Code § 22-2401 (1973).
The rape statute under consideration reads, in relevant part:
“Whoever has carnal knowledge of a female forcibly and against her will . . . shall be imprisoned for any term of years or for life.” D. C. Code § 22-2801 (1973).
If one tests the above-quoted statutes in the abstract, one can see that rape is not a lesser included offense of felony murder, because proof of the latter will not necessarily require proof of the former. One can commit felony murder without rape and one can rape without committing felony murder. If one chooses to apply Blockburger to the indictment in the present case, however, rape is a “lesser included offense” of felony murder because,- in this particular case, the prosecution could not prove felony murder without proving the predicate rape.
Because this Court has. never been forced to apply Blockburger in the context of compound and predicate offenses,5 *711we have not had to decide whether Blockburger should be applied abstractly to the statutes in question or specifically to the indictment as framed in a particular case. Our past decisions seem to have assumed, however, that Blockburger’s analysis stands or falls on the wording of the statutes alone. Thus, in Blockburger itself the Court stated that “the applicable rule is that where the same act or transaction constitutes a violation of two distinct statutory provisions, the test to be applied to determine, whether there are two offenses or only one, is whether each provision requires proof of a fact which the other does not.” 284 U. S., at 304 (emphasis added). More recently, we framed the test as whether “ 'each statute requires proof of an additional fact which the other does not. . . Brown v. Ohio, supra, at 166, quoting Morey v. Commonwealth, 108 Mass. 433, 434 (1871) (emphasis added). See also Iannelli v. United States, 420 U. S., at 785, n. 17 (“[T]he Court’s application of the [Blockburger] test focuses on the statutory elements of the offense”); M. Fried-land, Double Jeopardy 212-213 (1969) (noting the two possible interpretations and pointing out that “the word ‘provision’ is specifically used in the test” as stated in Blockburger). Moreover, because the Blockburger test is simply an attempt to determine legislative intent, it seems more natural to apply it to the language as drafted by the legislature than to the wording of a particular indictment.
The Court notes this ambiguity but chooses instead to apply the test to the indictment in the present case.6 See *712ante, at 693-694. In doing so, it offers only two reasons for rejecting what would seem to be the more plausible interpretation of Blockburger. First, the Court notes that Congress could have broken felony murder down in six separate statutory provisions, one for each of the predicate offenses specified in § 22-2401, thereby insuring that, under Block-burger, rape would be a lesser included offense of murder in the course of rape. According to the Court, “[i]t is doubtful that Congress could have imagined that so formal a difference in drafting had any practical significance, and we ascribe none to it.” Ante, at 694. The short answer to this argument is that Congress did not break felony murder down into six separate statutory provisions. Thus, it hardly avails the Court to apply Blockburger to a statute that Congress did not enact. More significantly, however, I believe that the Court’s example illustrates one of my central points: when applied to-compound and predicate offenses, the Blockburger test has nothing whatsoever to do with legislative intent, turning instead on arbitrary assumptions and syntactical subtleties. Cf. n. 6, supra. If the polestar in this case is to be legislative intent, I see no reason to apply Blockburger unless it advances that inquiry.
Second, the Court asserts that “to the extent that . . . the matter is not entirely free of doubt, the doubt must be re*713solved in favor of lenity.” Ante, at 694. This assertion, I would suggest, forms the real foundation of the Court’s decision. Finding no indication in the legislative history whether Congress intended cumulative punishment, and applying Blockburger with insolubly ambiguous results, the Court simply resolves its doubts in favor of petitioner and concludes that the rape committed by petitioner must merge into his conviction for felony murder. In doing so, the Court neglects the one source that should have been the starting point for its entire analysis: the lower court’s construction of the relevant statutes.
Unlike this Court, the District of Columbia Court of Appeals looked beyond the ambiguous legislative history and the inconclusive Blockburger test to examine the common-law roots of the crime of felony murder and to consider the societal interests protected by the relevant statutes. As for the first source, the lower court concluded from the history of felony murder at common law that “while the underlying felony is an element of felony murder it serves a more important function as an intent-divining mechanism” and that merger of the two offenses was therefore “inappropriate.” 379 A. 2d 1152, 1160 (1977). In so reasoning, the lower court acted in conformity with this Court’s long tradition of reading criminal statutes enacted by Congress “in the light of the common law. . . .” United States v. Carll, 105 U. S. 611, 612 (1882). See also Morissette v. United States, 342 U. S. 246, 262-263 (1952).
In addition to looking to the common law for assistance in determining Congress’ intent, the lower court examined “the societal interests protected by the statutes under consideration.” 379 A. 2d, at 1158-1159. Because § 22-2801 was designed “to protect women from sexual assault” while § 22-2401 was intended “to protect human life,” the court concluded that cumulative punishment was permissible. 379 A. 2d, at 1159. Indeed, the Blockburger test itself could be *714viewed as nothing but a rough proxy for such analysis, since, by asking whether two separate statutes each include an element the other does not, a court is really asking whether the legislature manifested an intention to serve two different interests in enacting the two statutes.
Ill
In sum, I find the lower court's reliance upon articulated considerations much more persuasive than this Court’s capitulation to supposedly hopeless ambiguity. But even if the case were closer, I do not see how the lower court’s conclusion could be classified as “egregious error” so as to justify our superimposing our own - admittedly dubious construction of the statutes in question on the District of Columbia. Unless we are going to forgo deference to the interpretation of the highest court of the District of Columbia on matters of local applicability and are going to push several other well-recognized principles of statutory and constitutional construction out of shape, with consequences for the federal system for the 50 States, I would hope that the Court’s decision would be one ultimately based on the “rule of lenity.” Because I believe that the question confronting us is purely one of statutory construction and because I believe the analysis indulged in by the Court of Appeals for the District of Columbia comes far closer to the proper ascertainment of congressional intent than does this Court’s opinion, I would affirm the judgment of the District of Columbia Court of Appeals.
The “same evidence” test was first formulated in Morey v. Commonwealth, 108 Mass. 433, 434 (1871), where the Supreme Judicial Court of Massachusetts held:
“A conviction or acquittal upon one indictment is no bar to a subsequent conviction and sentence upon another, unless the evidence required to support a conviction upon one of them would have been sufficient to warrant a conviction upon the other. The test is not whether the defendant has already been tried for the same act, but whether he has been put in jeopardy for the same offence. A single act may be an offence against two statutes; and if each statute requires proof of an additional fact which the other does not, an acquittal or conviction under either statute does not exempt the defendant from prosecution and punishment under the other.”
This Court has placed varying degrees of reliance upon this test both in the context of successive prosecutions, see, e. g., Brown v. Ohio, 432 U. S. 161 (1977); Gavieres v. United States, 220 U. S. 338, 342 (1911), and in the context of multiple punishments imposed at a single criminal proceeding. See, e. g., Blockburger v. United States, 284 U. S. 299 (1932); Carter v. McClaughry, 183 U. S. 365 (1902). See also infra, at 707-714.
We are not dealing here, of course, with a case where a state court has engaged in “retroactive lawmaking” by interpreting a local statute in an unforeseeable manner. Compare Bouie v. City of Columbia, 378 U. S. 347 (1964), with Rose v. Locke, 423 U. S. 48 (1975).
It should not matter whether the Blockburger test enters this case as a common canon of statutory construction, see Iannelli v. United States, 420 U. S. 770, 785, n. 17 (1975), or through the “less than felicitous” phrasing of D. C. Code § 23-112. See ante, at 691 (opinion of the Court). In either case, the dispositive question is whether the legislature intended to allow multiple punishments, and the Blockburger test should be employed only to the extent that it advances that inquiry.
In this regard, see also the discussion of the sentencing scheme under 18 U. S. C. § 924 (c) (1), infra, at 709.
But see Simpson v. United States, 435 U. S. 6, 11-12, and n. 6 (1978) (reserving application of Blockburger in context of 18 U. S. C. § 924 (c)); Jeffers v. United States, 432 U. S. 137, 149-150 (1977) Blackmun, J.) (assuming, arguendo, that 21 U. S. C. § 846 is a lesser *711included offense of 21 U. S. C. § 848). But see also American Tobacco Co. v. United States, 328 U. S. 781, 788 (1946) (finding, under Blocburger, that conspiracies to violate §§ 1 and 2 of the Sherman Act could be punished separately).
The Court denies that it applies the Blockburger test to the indictment in this case, asserting instead that it merely concludes that “rape [is] to be considered a lesser offense included within the offense of a killing in the course of rape.” Ante, at 694, n. 8. Our disagreement on this matter turns on the elusive meaning of the word “offense.” Technically, *712§ 22-2401 defines only one offense, murder in the first degree, which can be committed in any number of ways. Even if the inquiry is limited to the “sub-offense” of felony murder, § 22-2401 indicates that a person may be convicted if he kills purposely in the course of committing any felony or kills even accidentally in the course of committing one of six specified felonies. Only by limiting the inquiry to a killing committed in the course of a rape, a feat that cannot be accomplished without reference to the facts alleged in -this particular case, can the Court conclude that the predicate offense is necessarily included in the compound offense under Blockburger. Because this Court has never before had to apply the Blockburger test to compound and predicate offenses, see n. 5, supra, and accompanying text, there is simply no precedent for parsing a single statutory provision in this fashion.