The defendant, an accountant who worked in Illinois, embezzled funds of his employer by writing, without authorization, three checks to himself aggregating more than $135,000 during a two-month period and depositing them in his personal account in a Maryland bank that happens to have offices only in that state and in Virginia. The locations of his place of work and of his bank are relevant because he was charged not only with willful failure to report his embezzled income on his federal income tax return for the year in which he received that income, 26 U.S.C. sec. 7206(1), but also with having transported money obtained by fraud across state lines. 18 U.S.C. sec. 2314. He was convicted of both crimes, was sentenced to a total of 27 months in prison, and appeals, challenging only the sufficiency of the evidence to convict him.

The appeal bespeaks a deep or perhaps desperate misunderstanding of the law of evidence, especially with regard to the second charge. It is conceded that all the government had to show was that the defendant had caused the checks to be transported across state lines. It is also conceded that the checks indeed ended up in either a Maryland or a Virginia office of the bank in which they were deposited and that one of the checks was accompanied by a note to the bank in the defendant's handwriting telling the bank what to do with the check. Presumably the defendant mailed the checks (and the accompanying note) to the bank but there is no direct evidence of this; that is, no one testified to having seen the defendant mail these items and he does not admit having mailed them. He argues that therefore he cannot be proved guilty beyond a reasonable doubt of having caused them to be transported across state lines.

He admits that certainty is not required to establish guilt beyond a reasonable doubt, that a conviction based on fingerprint evidence for example cannot be overturned by pointing out that there is some minute probability of erroneous fingerprint identification. But he insists that a conviction can survive remote doubts about its correctness only when it is based on computed probabilities; common sense probabilities will not do. That is wrong. Common sense as well as science is a source of justified true beliefs, including warranted confidence that certain probabilities though unquantified are so slight that they do not create reasonable doubt. Often this confidence is formed by comparing hypotheses and sensibly adjudging one to be vastly more probable than the others, even taken all together. This case illustrates that routine and unexceptionable reasoning process nicely. While it is conceivable that the defendant did not cause the checks to end up in Maryland or Virginia--maybe after writing them and the accompanying note he changed his mind and threw them in the wastepaper basket in his office and the cleaning people picked them up after hours and mailed them to the defendant's bank--this hypothesis is so unlikely to be true that in the absence of any evidence in support of it (and there is none) a rational jury would be entirely justified in dismissing the probability of its being true as minute in relation to the probability that the defendant himself caused the checks to end up where they did; indeed a jury would be irrational to conclude otherwise.

The situation is a little more doubtful with respect to the defendant's conviction for willfully filing a false return. To be convicted of that offense he had to be proved to have known (at least if he made an issue of his knowledge) that illegal income is taxable and so his income from embezzlement should have been reported, 26 U.S.C. sec. 7206(1); Cheek v. United States, 498 U.S. 192, 201-02 (1991); United States v. Kontny, 238 F.3d 815, 820 (7th Cir. 2001); United States v. Peters, 153 F.3d 445, 461 (7th Cir. 1998); United States v. Pirro, 212 F.3d 86, 89 (2d Cir. 2000), since otherwise the return though false would not be willfully so; and again there is no direct evidence that he knew- -evidence that could only have taken the form of an admission by him, for that is the only form that direct evidence of a person's state of mind can take. (There are no eyewitnesses to the mental contents of a person's head, as opposed to external phenomena, such as the placing of a letter in a mailbox.) But the circumstantial evidence was convincing, and the absence of direct evidence therefore no bar to conviction. E.g., Spies v. United States, 317 U.S. 492, 499-500 (1943); United States v. Paneras, 222 F.3d 406, 410 (7th Cir. 2000); United States v. Robinson, 177 F.3d 643, 648 (7th Cir. 1999); United States v. Townsend, 924 F.2d 1385, 1390 (7th Cir. 1991); United States v. Radtke, 799 F.2d 298, 302 (7th Cir. 1986); United States v. Guidry, 199 F.3d 1150, 1156-58 (10th Cir. 1999). We are, however, doubtful about the validity of the government's argument, based on the case just cited, id. at 1157, that the defendant's efforts (which incidentally were rather feeble) to conceal the embezzlement was evidence that he knew that embezzled income is taxable. For he would have had an incentive to conceal the embezzlement in order to save his job and avoid prosecution for embezzlement, even if embezzled income were tax-free. But there is plenty of other circumstantial evidence of the defendant's intent to avoid tax. As in Guidry, the defendant was an accountant, moreover an experienced one; he personally prepared the fraudulent tax return; the sums taken were large (they amounted to 75 percent of his total income during the period of the embezzlement), and (the weakest bit of evidence) he used the money for ordinary expenses, the sort of thing people usually defray from taxable income. Furthermore, the fact that illegal income is taxable is widely known, even among lay people. Everyone knows that Al Capone, for example, was nailed for income-tax evasion, not for the bootlegging, loan-sharking, extortion, and prostitution that generated the income. Accountants know better than anyone except tax lawyers that illegal income is taxable.

It is possible nevertheless that this experienced accountant who prepared his own tax return thought that illegal income was tax exempt, but it is too remote a possibility to compel an acquittal, at least in the absence of any evidence that might make it plausible, such as that the defendant suffered from some psychiatric disorder that had deranged his knowledge of elementary tax law or his ability to use that knowledge in filling out his income tax return. Anything is possible; there are no metaphysical certainties accessible to human reason; but a merely metaphysical doubt (for example, doubt whether the external world is real, rather than being merely a dream) is not a reasonable doubt for purposes of the criminal law. See, e.g., Victor v. Nebraska, 511 U.S. 1, 13- 17 (1994) ("absolute certainty is unattainable in matters relating to human affairs," id. at 13); United States v. Williams, 216 F.3d 1099, 1103-04 (D.C. Cir. 2000); United States v. Guidry, supra, 199 F.3d at 1157-58; United States v. Delpit, 94 F.3d 1134, 1148 (8th Cir. 1996); United States v. Hall, 854 F.2d 1036, 1044 (7th Cir. 1988) (concurring opinion); see also Holland v. United States, 348 U.S. 121, 139-40 (1954). If it were, no one could be convicted. As with the transportation of the checks, so with the charge of willfully filing a false return, the jury had to choose between two hypotheses. One was that the defendant knew that embezzled income is taxable. The other was that he thought embezzled income tax-free--a token of the government's affection for embezzlers and other thieves. The former hypothesis was, in the circumstances, far more likely than the latter. That is enough to compel affirmance.

Affirmed.