with whom Justice Thomas joins, concurring in the judgment.
AlS the proliferation of opinions indicates, this is a most difficult case. I agree with the Court’s conclusion that the Assimilative Crimes Act (ACA), 18 U. S. C. § 13(a), does not incorporate Louisiana’s first-degree murder statute into the criminal law governing federal enclaves in that State. I write separately because it seems to me that the Court’s manner of reaching that result turns the language of the ACA into an empty vessel, and invites the lower courts to fill it with free-ranging speculation about the result that Congress would prefer in each case. Although I agree that the ACA is not a model of legislative draftsmanship, I believe we have an obligation to search harder for its meaning before abandoning the field to judicial intuition.
The Court quotes the text of the ACA early in its opinion, but then identifies several policy reasons for leaving it behind. The statutory language is deceptively simple.
*174“Whoever within or upon any [federal enclave] is guilty of any act or omission which, although not made punishable by any enactment of Congress, would be punishable if committed or omitted within the jurisdiction of the State ... in which such place is situated, . . . shall be guilty of a like offense and subject to a like punishment.” § 13(a).
At first glance, this appears to say that state law is not assimilated if the defendant can be prosecuted under any federal statute. The Court acknowledges this, but concludes that “a literal reading of the words ‘any enactment’ would dramatically separate the statute from its intended purpose,” ante, at 160, because, for example, a general federal assault statute would prevent assimilation of a state prohibition against murder.
It seems to me that the term “any enactment” is not the text that poses the difficulty. Whether a federal assault statute (which is assuredly an “enactment”) prevents assimilation of a state murder statute to punish an assault that results in death depends principally upon whether fatal assault constitutes the same “act or omission” that the assault statute punishes. Many hypothetieals posing the same issue can readily be conceived of. For example, whether a state murder statute is barred from assimilation by a federal double-parking prohibition, when the behavior in question consists of the defendant’s stopping and jumping out of his car in the traffic lane to assault and kill the victim. The federal parking prohibition is sure enough an “enactment,” but the issue is whether the “act or omission” to which it applies is a different one. So also with a federal statute punishing insurance fraud, where the murderer kills in order to collect a life insurance policy on the victim.
Many lower courts have analyzed situations like these under what they call the “precise acts” test, see, e. g., United States v. Kaufman, 862 F. 2d 236 (CA9 1988), which in practice is no test at all but an appeal to vague policy intuitions. *175See, e. g., United States v. Brown, 608 E 2d 551 (CA5 1979) (striking a child is not the same “precise act” for purposes of a federal assault law and a state law against child abuse). I am skeptical of any interpretation which leaves a statute doing no real interpretive work in most of the hard cases which it was drafted to resolve. On that score, however, the Court’s solution is no improvement. After rejecting proposals from the petitioner and from the United States that would have given the ACA more definite content (on the policy grounds that they would produce too little, and too much, assimilation, respectively), the Court invites judges to speculate about whether Congress would approve of assimilation in each particular case.
“[T]he court must ask ... whether the federal statutes that apply to the ‘act or omission’ preclude application of the state law in question, say, because its application would interfere with the achievement of a federal policy, because the state law would effectively rewrite an offense definition that Congress carefully considered, or because federal statutes reveal an intent to occupy so much of a field as would exclude use of the particular state statute at issue .... The primary question (we repeat) is one of legislative intent: Does applicable federal law indicate an intent to punish conduct such as the defendant’s to the exclusion of the particular state statute at issue?” Ante, at 164, 166 (citations omitted).
Those questions simply transform the ACA into a mirror that reflects the judge’s assessment of whether assimilation of a particular state law would be good federal policy.
I believe that the statutory history of the ACA supports a more principled and constraining interpretation of the current language. The original version of the ACA provided for assimilation whenever “any offence shall be committed ..., the punishment of which offence is not specially provided for by any law of the United States.” 4 Stat. 115. *176Subsequent amendments replaced the word “offence” with “act or thing,” 35 Stat. 1145, and eventually the present formulation, “act or omission.” But we held in Williams v. United States, 327 U. S. 711, 722-723 (1946), that those amendments were designed to respond to a perceived technical deficiency, and that they did not intend to change the meaning of the Act.
Williams reached that conclusion by studying the legislative history of the ACA amendments. Although I am not prepared to endorse that particular methodology, reading the ACA against the backdrop of its statutory predecessors does shed some light on its otherwise puzzling language. An “act or omission . . . made punishable by [law]” is the very definition of a criminal “offense,” and certainly might have been another way to express that same idea. In addition, the ACA still provides that a defendant charged with an assimilated state crime “shall be guilty of a like offense and subject to a like punishment.” 18 U. S. C. § 13(a) (emphasis added). Since an interpretation that ascribes greater substantive significance to the amendments would produce such a vague and unhelpful statute, I think that Williams’s reading of the ACA was essentially correct. A defendant may therefore be prosecuted under the ACA for an “offense” which is “like” the one defined by state law if, and only if, that same “offense” is not also defined by federal law.
That interpretation would hardly dispel all of the confusion surrounding the ACA, because courts would still have to decide whether the assimilated state offense is “the same” as some crime defined by federal law. As Justice Kennedy points out in dissent, “[t]here is a methodology at hand for this purpose, and it is the Blockburger test we use in double jeopardy law.” Post, at 182. Two offenses are different, for double jeopardy purposes, whenever each contains an element that the other does not. See, e. g., Blockburger v. United States, 284 U. S. 299, 304 (1932). That test can be *177easily and mechanically applied, and has the virtue of producing consistent and predictable results.
The Blockburger test, however, establishes what constitutes the “same offence” for purposes of the traditional practice that underlies the Double Jeopardy Clause, U. S. Const., Arndt. 5. That constitutional guarantee not only assumes a scheme of “offences” much more orderly than those referred to by the ACA (since they are the offenses designed by a single sovereign), but also pursues policy concerns that are entirely different. When it is fair to try a defendant a second time has little to do with when it is desirable to subject a defendant to two separate criminal prohibitions. Thus, for example, double jeopardy law treats greater and lesser included offenses as the same, see, e. g., Harris v. Oklahoma, 433 U. S. 682 (1977) (per curiam), so that a person tried for felony murder cannot subsequently be prosecuted for the armed robbery that constituted the charged felony. That is fair enough; but it is assuredly not desirable that a jurisdiction (the federal enclave) which has an armed robbery law not have a felony-murder law. Contrariwise, as the Court’s opinion points out, ante, at 163, Blockburger’s, emphasis on the formal elements of crimes causes it to deny the “sameness” of some quite similar offenses because of trivial differences in the way they are defined. In other words, the Blockburger test gives the phrase “same offence” a technical meaning that reflects our double jeopardy traditions, see Grady v. Corbin, 495 U. S. 508, 528-536 (1990) (Scalia, J., dissenting), but that is neither a layman’s understanding of the term nor a meaning that produces sensible results for purposes of “gap filling.” There is no reason to assume, it seems to me, that Congress had the term of art in the Double Jeopardy Clause in mind when it enacted the ACA.
Justice Kennedy contends that all of these concerns can be accommodated through adjustments to the Blockburger test. In his view, for example, “the existence of a lesser included federal offense does not prevent the assimilation of *178a greater state offense under the ACA, or vice versa.” Post, at 183. He proposes that courts should “look beyond slight differences in wording and jurisdictional elements to discern whether, as a practical matter, the elements of the two crimes are the same.” Post, at 182. In order to avoid overruling Williams, he also suggests that assimilation is improper when “Congress .. . adverts to a specific element of an offense and sets it at a level different from the level set by state law.” Post, at 183. I admire Justice Kennedy’s effort to construct an interpretation of the ACA that yields more certain and predictable results, but the modifications he proposes largely dispel the virtues of familiarity, clarity, and predictability that would make Blockburger the means to such an end. Ultimately, moreover, those modifications are driven by a view of the policies underlying the Act which I do not share. Justice Kennedy contends that the ACA is primarily about federalism, and that respect for that principle requires a strong presumption in favor of assimilation. Post, at 181-182. To the extent that we can divine anything about the ACA’s “purpose” from the historical context which produced it, I agree with the Court that the statute was apparently designed “to fill gaps in the federal criminal law” at a time when there was almost no federal criminal law. Ante, at 160; see also Williams, supra, at 718-719.
Rejecting Blockburger's elements test leaves me without an easy and mechanical answer to the question of when a state and federal offense are the “same” under the ACA. But the language of the original 1825 ACA suggests that the focus of that inquiry should be on the way that crimes were traditionally defined and categorized at common law. It provided that
“... if any offence shall be committed in [an enclave], the punishment of which offence is not specially provided for by any law of the United States, such offence shall.. . receive the same punishment as the laws of the state... *179provide for the like offence when committed within the body of any county of such state.” 4 Stat. 115.
Congress did not provide any methodology for determining whether an “offence” under state law is “provided for by any law of the United States”; the statute appears, instead, to presume the reader’s familiarity with a set of discrete “of-fence[s]” existing apart from the particular provisions of either state or federal statutory law.
In my opinion, the legal community of that day could only have regarded such language as a reference to the traditional vocabulary and categories of the common law. Indeed, the original ACA was at least in part a response to our decision in United States v. Hudson, 7 Cranch 32 (1812), which held that the federal courts could not recognize and punish common-law crimes in the absence of a specific federal statute. The common law’s taxonomy of criminal behavior developed over the centuries through the interplay of statutes and judicial decisions, and its basic categories of criminal offenses remain familiar today: murder, rape, assault, burglary, larceny, fraud, forgery, and so on. I believe that a contemporary reader of the original ACA would have understood it to apply if, and only if, the federal criminal statutes simply failed to cover some significant “offence” category generally understood to be part of the common law.
Since 1825, of course, state and federal legislatures have created a tremendous variety of new statutory crimes that both cut across and expand the old common-law categories. Some of those new “offences” may have become so well established in our common legal culture that their absence from the federal criminal law would now represent a significant gap in its coverage — a gap of the sort the ACA was designed to fill. That possibility introduces an unavoidable element of judgment and discretion into the application of the ACA, and to that extent my interpretation is subject to the same criticisms I have leveled at the approaches taken by the Court and by Justice Kennedy. But I think that *180danger is more theoretical than practical. The structure of the criminal law, like the basic categories of human vice, has remained quite stable over the centuries. There have been a few genuine innovations recently; I have in mind, for example, antitrust or securities crimes which did not exist in 1825. But Congress has been the principal innovator in most of those areas, and I doubt that courts will confront many new “offence” candidates that are not already covered by the federal criminal law. Regardless, the approach outlined above would produce more predictable results than the majority’s balancing test, and has the additional virtue of being more firmly grounded in the text and statutory history.
It also produces a clear answer in this case. Ms. Lewis’s conduct is not just punishable under some federal criminal statute; it is punishable as murder under 18 U. S. C. § 1111. Louisiana’s murder statutes are structured somewhat differently from them federal counterparts, but they are still unquestionably murder statutes. Because that “offence” is certainly “made punishable by any enactment of Congress,” there is no gap for the ACA to fill. That remains true even if the common-law category at the appropriate level of generality is instead murder in the first degree. That “offence” is also defined and punished by the federal criminal law, although the prosecutors in this ease apparently did not believe that they could establish its elements. Accordingly, I concur in the judgment, and in Part IY of the majority’s opinion.