concurring and dissenting.
I agree with the majority’s analysis and conclusions in all respects but one. As to Count XI of the indictment, which charged Stansfield with witness tampering, I cannot agree with the majority’s conclusion that “the evidence was sufficient to support a conviction under 18 U.S.C. § 1512(a)(1)(C).” Maj. Op. at 919. In my view, this result rests upon an erroneous application of 18 U.S.C. § 1512(a)(1)(C), one which is not supported by either the case law on which the majority relies or the language of the statute itself. And our disagreement is significant not only because of the effect the majority’s application will have on the government’s burden of proof in future cases brought under this statute, but because if the majority is correct Stansfield will in all likelihood be retried on this twenty-year count, whereas if my understanding is correct the appropriate disposition is a judgment of acquittal due to the government’s failure to introduce evidence to sustain a conviction on this count.
The majority purportedly rejects the government’s argument that it “need only prove that the offense about which the defendant wishes to prevent communications is actually a federal offense.” Maj. Op. at 918 (emphasis added). Here, the majority and I are in agreement, for the government’s argument essentially asks that it be required to prove nothing. Instead, the majority offers a four-part construction of the statute, which requires the government to prove, inter alia, that the defendant believed that the victim might communicate with federal authorities. Id. at 918. Unfortunately, the majority fails to apply its own construction of the statute, finding Stansfield’s conviction supportable based solely on the federal nature of his offense and evidence showing that a file had been opened by federal authorities. Stans-field’s intent is irrelevant under the majority’s analysis. Contending that Stansfield’s intent “may be inferred,” id., from the mere fact that the underlying offense happened to be a federal offense, even though the government offered absolutely no evidence to show that Stansfield had any knowledge, awareness or belief that he had committed a feder*923al offense or was the subject of a federal investigation, the majority essentially eviscerates the intent element of the statute. I cannot agree with this analysis. Instead, I believe the government must demonstrate, through direct or circumstantial evidence— but beyond a reasonable doubt — that Stans-field believed that the underlying offense was federal or was being investigated by federal authorities with whom the victim of the threats might communicate. It is this important element — some evidence of the defendant’s awareness of the federal nature of his crime, which in turn could form the basis for his intent to prevent a communication about it to a federal official — that is missing from both the majority’s analysis and the government’s proof in this case. And while the majority concedes that proof of either of these things would be sufficient to establish a defendant’s intent, it offers no explanation as to how else a defendant’s intent could possibly be established.1
Moreover, the majority’s reliance on case law from other circuits to support its conclusion is misguided. The question presented in Romero, Edwards, and Leisure was whether § 1512 required the presence of an ongoing federal investigation to support a conviction for witness tampering under the statute. See United States v. Romero, 54 F.3d 56 (2d Cir.1995), cert. denied, — U.S. -, 116 S.Ct. 1449, 134 L.Ed.2d 568 (1996); United States v. Edwards, 36 F.3d 639 (7th Cir.1994); United States v. Leisure, 844 F.2d 1347 (8th Cir.1988). While it is true that these cases rejected the requirement of an ongoing investigation, they do not support the majority’s conclusion here that a defendant’s intent can be inferred, without more, by the mere fact that the defendant committed a federal crime. On the contrary, in each of these cases the defendant had knowledge of the federal nature of the crime and of the consequent potential of a federal investigation.2 In Romero, for example, the court noted that “members of [the defendant’s] organization had become suspicious that [the witness] was cooperating with federal authorities.” Romero, 54 F.3d at 59. And in Edwards, the government presented testimony which indicated that the defendants killed the victim “because they feared he was informing the DEA about their operations.” Edwards, 36 F.3d at 645 (emphasis added). Thus, although the government need not *924prove the presence of an ongoing federal investigation, it must at least prove “that the defendant believed that a person might furnish information to federal officials and that he killed or attempted to kill that person in order to prevent such disclosure.” Id. (emphasis added) (citing Leisure, 844 F.2d at 1364).
Title 18 U.S.C. § 1512(a)(1)(C) provides: “Whoever kills or attempts to kill another person with intent to prevent the communication by any person to a law enforcement officer ... of the United States of information relating to the commission or possible commission of a Federal offense” shall be imprisoned for up to twenty years. As the majority notes, “law enforcement officer” is defined as “an officer or employee of the Federal government, or a person authorized to act for or on behalf of the Federal government....” 18 U.S.C. § 1515(a)(4)(A). Thus, at a minimum, the government must prove that the defendant believed that there was a possibility of a federal investigation.
The majority is overly generous to the government, however, in stating that Stans-field “was aware that some investigation, though not necessarily a federal one, was under way,” and concluding that this is enough to support his conviction. Maj. Op. at 919. . In fact, the government did not introduce any evidence to suggest that Stansfield could have known of the possibility of a, federal investigation. The evidence only established that Stansfield knew that the Pennsylvania State Police and Erie Insurance Company were interviewing potential witnesses and pursuing an investigation and that Hoffman had been interviewed by both. It is obvious from the evidence that he was angry that Hoffman might be cooperating with a state official (who happened to be the only law enforcement officer conducting any investigation at that time), not a federal official. We don’t know, and the government has not shown, that he even had any reason to believe that his offenses were federal in nature, or that the victims might communicate with federal authorities. In light of the total lack of evidence in this respect, I am perplexed by the majority’s conclusion that a jury could have concluded beyond a reasonable doubt that Stansfield believed that Hoffman might cooperate with federal authorities. There is simply no basis in the record to justify this conclusion.
Moreover, while the majority states that “the evidence also showed that federal authorities had begun an investigation approximately one month prior to the conduct in question[,]” id. at 919, in fact, the evidence showed that Postal Inspectors had merely opened a file on the case and had done nothing more. But unless he had a mole inside the Postal Inspection Service or was clairvoyant, Stansfield surely was not aware that a file had been opened. And since nothing else of a federal investigative nature had occurred, Stansfield could not have known that a federal investigation (if that’s what opening a file is) was in the works.
On the record before us, therefore, there is simply no way to conclude that Stansfield either believed that a federal investigation was underway or could possibly have been aware of the potential for a federal investigation. While it is easy for those of us versed in the federal law to conclude, as the majority does, that “the underlying offense clearly was a federal offense[,]” id. at 919, Stansfield is neither a lawyer nor a judge, and the government has yet to demonstrate, even remotely, that Stansfield had reason to believe that the underlying offense was federal in nature, or had reason to believe that Hoffman might have talked to “an officer or employee of the federal government” (which he had not), or might do so in the" future. He could not possibly have intended to frustrate a communication he had no reason to believe might occur, to a person he had no reason to believe existed.
Despite the majority’s protestations to the contrary, the effect of the majority’s construction and application of this important statute is not only to diminish, but as a practical matter to remove, the government’s burden of proof. Unlike the eases on which the majority relies, here the government did not prove that Stansfield believed or was aware of the possibility or existence of a federal investigation, yet the majority finds that lack of proof irrelevant because, once again, it can be “inferred” from the fact that *925Stansfield’s underlying offense just happened to be a federal offense. To relieve the government of its burden of proving an important element of a criminal offense is a serious step, and I believe the majority’s approach here is seriously out of step with both the case law and the statute.
I would enter a judgment of acquittal on Count XI. I concur in the remainder of the majority opinion.
. The majority asserts that I would “impose on the government the burden of proving that the defendant knew or believed either that the offense was federal in nature or that there was a federal investigation in progress.” Maj. Op. at 918 n.4. While this is true, the majority misunderstands my point. In doing so, it fails to explain how anything short of such proof could give rise to a reasonable conclusion that a defendant intended to frustrate a communication to a federal official, which is what the statute requires. Of course, that is impossible to explain because there is no way a defendant's intent can be established without proof of one or the other. Such proof is non-existent in this case. Thus, my larger point is that the government made no suggestion that Stansfield knew or believed, or even had any reason to know or believe, that any offense he committed was a federal offense. Nor did the government offer any proof to show that Stansfield knew of or believed that a federal investigation was underway. In any event, my concern here is not how Stansfield's intent is proven, but that it is proven.
. The majority concludes that this fact is not relevant to our analysis because the courts in Edwards, Romero, and leisure did not explicitly state that the defendant's knowledge or belief in the federal nature of the offense or the defendant's knowledge or belief in the possibility of a federal investigation was central to their decisions. In those cases, however, each defendant argued that the government must prove the presence of an ongoing investigation. Rejecting that contention, the courts reviewed the government's evidence and held, respectively, that the government had offered enough proof for a reasonable juror to conclude that the defendant possessed the requisite intent. Thus, one can learn from a close reading of those cases that the government at least presented some testimony that would allow a jury to infer that the defendant intended to prevent the victim from conveying information to federal authorities, as opposed to some general, undefined law enforcement authority. See Leisure, 844 F.2d at 1364. Here, however, the government presented no such testimony. The majority glosses over this evidentiary void in determining that Stansfield's knowledge may simply be “inferred” from the federal nature of the underlying offense. But when a statute requires a specific showing, proof in a vacuum is tantamount to no proof at all. In this case, the majority, which reverses on other grounds, would nonetheless sustain the defendant’s conviction for witness tampering, which carries a twenty-year sentence, not on the basis of evidence presented in court, but on some vague, remote, unfocused, ill-defined “inference.”