Opinion for the Court filed by Circuit Judge WILLIAMS.

STEPHEN F. WILLIAMS, Circuit Judge:

Andrew L. Stone and Francis N. Rosen-baum appeal from judgments of the Tax Court finding several million dollars of tax and penalties due on income allegedly received in 1963-67. (Stone’s wife, M. Jeanne Stone, also appeals from a judgment imposing liability on her as recipient of an allegedly fraudulent transfer from Stone, but under our disposition her contentions are moot. See pages 351-352 below.) In the years at issue, Stone and Rosen-baum were the principal figures in Chrom-craft Corporation, a supplier of rocket launchers to the United States armed forces. Stone was president; he also owned 75% of the stock until March 2, 1965, when he became sole shareholder. Rosenbaum was a director and tax counsel. In their positions in Chromcraft they en *343 gaged in a series of complex transactions that we will assume were designed to mulct the United States government, either as a purchaser of defense equipment, or as collector of taxes due from Chromcraft, or both. Indeed, both Stone and Rosenbaum pleaded guilty to conspiracy to commit an offense or to defraud the United States, 18 U.S.C. § 371 (1982), and to eight counts of making knowingly false, fictitious or fraudulent statements or representations to the Government, 18 U.S.C. § 1001 (1982), based on those transactions, see Alsco-Harvard Fraud Litigation, 523 F.Supp. 790, 800 (D.D.C.1981), and served prison sentences. The transactions involved payments by Chromcraft to intermediaries, usually its subcontractors or suppliers, and then payments by the latter nominally to Chrom-craft or to “dummy” Swiss entities whose sole role in the transactions was to disguise the destination of the payments. The intermediaries’ payments, aggregating about $4 million, Joint Appendix (“J.A.”) 182, were deposited in a series of Swiss bank accounts, known here as the Swiss Funds. Rosenbaum exercised direct control over all these funds, as a signatory on all the accounts; Stone, a signatory on some accounts but not on others, also exercised some control, largely if not wholly indirect. Here we deal solely with the Commissioner’s contention that for income tax purposes Chromcraft’s payments to the Swiss Funds via the intermediaries were income to Stone and Rosenbaum personally — to Stone as a constructive dividend from Chromcraft, to Rosenbaum simply as unreported income.

The parties do not dispute the essential substantive law governing these transactions, although the formulas on which they agree are somewhat elusive. They agree, for example, on the generality that if Stone and Rosenbaum received an economic benefit from the Swiss Funds, they are liable for tax on at least whatever benefit they received. See, e.g., Rutkin v. United States, 343 U.S. 130, 137, 72 S.Ct. 571, 575, 96 L.Ed. 833 (1952) (a cash receipt under circumstances “allowpng] the recipient freedom to dispose of it at will [is income], even though it may have been obtained by fraud and his freedom to use it may be assailable by someone with a better title to it”); Dawkins v. Commissioner of Internal Revenue, 238 F.2d 174, 178 (8th Cir.1956) (citing Rutkin). They recognize that in contexts such as assignments of income, the “power to dispose of income” will be benefit enough to trigger taxation. Helvering v. Horst, 311 U.S. 112, 118, 61 S.Ct. 144, 147, 85 L.Ed. 75 (1940).

On the other hand, the parties also agree that even though control over funds is an economic benefit in the general sense of that term, the exercise of such control by a corporate officer as such does not constitute a benefit as the term is used in the generality invoked above. Thus, even when a corporate officer is its sole shareholder (and thus in ultimate control), and he transfers corporate funds to his personal checking account, and his dealings with the corporation are “extremely informal,” there is no constructive dividend so long as he can show that his intent “was to use such funds for corporate purposes as an agent of the corporation.” Nasser v. United States, 257 F.Supp. 443, 449 (N.D.Cal.1966). See also Loftin and Woodard, Inc. v. United States, 577 F.2d 1206, 1215 (5th Cir.1978) (constructive dividend is corporate payment primarily for shareholder’s benefit rather than for corporate benefit, determination of which is question of fact). Discussing corporate payments for the arguable benefit of shareholders, ranging from travel and entertainment to “payment of personal expenses in an aura of fraud,” Boris I. Bittker & James S. Eustice observe that the basic issue is “whether the corporate expenditure was incurred primarily to benefit the corporation’s trade or business, or primarily for the personal benefit of the shareholders.” See Federal Income Taxation of Corporations and Shareholders, 117.05(6) (5th Ed.1987). Bittker and Eustice point out that diversions of corporate funds have much in common with corporate payment of shareholder expenses, except that they may involve more concealment. Id.

The parties also appear to agree that these principles apply equally to Rosen-baum’s control over the Swiss Funds as a *344 corporate officer, though under the rubric of “unreported income” rather than constructive dividend.

Application of these standards is obviously difficult where corporate officers exercise command over secret Swiss bank accounts. The fact finder must infer the individuals’ intent in part from the objective evidence as to their actions, which may be ambiguous, and in part from their own testimony, which is almost by definition self-serving.

With the parties in agreement on the basic substantive law, the critical legal issue is the deference that the Tax Court owed the Special Trial Judge, to whom the matter was referred for fact-finding in May 1976. The Special Trial Judge rejected the Commissioner’s contention that Chrom-craft’s funds were income to Stone and Rosenbaum at the moments of transfer to the Swiss Funds. He concluded instead that the funds remained the property of Chromcraft. His report, filed February 1980, proceeded to a judge of the Tax Court designated to sit as a division of that Court. The Tax Court called no witnesses but reviewed the exhibits and 6000 pages of testimony accumulated by the Trial Judge. Declining to credit the testimony of Stone and Rosenbaum, it rejected the Trial Judge’s report and found Stone and Rosenbaum to have become the owners of the Swiss Funds (for tax purposes) at the moments of transfer.

In the face of these divergent fact findings, the scope of the Tax Court’s review of the Trial Judge is critical. We recognize that issues of scope of review often seem to have “no more substance at the core than a seedless grape.” Ernest Gellhorn & Glen Robinson, Perspectives on Administrative Law, 75 Colum.L.Rev. 771, 780 (1975). Nonetheless, we are persuaded that the language of the Tax Court Rule applicable to this case (and still applicable under a different number) sought to establish the relatively high level of deference that the phrase “clearly erroneous” entails. By contrast, the Tax Court described its review of the Trial Judge’s report in terms that suggest it saw the standard of review as falling somewhere between de novo review and a mild presumption in favor of the correctness of the Trial Judge’s decision. 1

We then turn to the factual details. Our review of the Tax Court undoubtedly proceeds under the clearly erroneous standard. See 26 U.S.C. § 7482(a)(1) (1982) (directing courts of appeals to review Tax Court decisions as they would those of district courts sitting without a jury); Fed.R.Civ.Pro. 52(a) (mandating use of clearly erroneous standard in review of district court fact-finding). Without resolving the exact meaning of such multi-layered deference, we find the Tax Court’s rejection of the Trial Judge’s findings to have been clearly erroneous in light of the deference that the Tax Court owed the Trial Judge. Accordingly we reverse and remand for computation of tax in accordance with the findings of the Trial Judge.

A Special Trial Judge—formerly called a “commissioner”—is an adjunct of the Tax Court, a judicial officer appointed pursuant to § 7456(c) of the Internal Revenue Code. See Tax Court Rule 180, 26 U.S.C.App. (1976), replaced, effective May 1979, by Tax Court Rule 180, 26 U.S.C.App. (1982). At the time the reference was made in the present case, the chief judge of the Tax Court was authorized to designate commissioners. See 26 U.S.C. §§ 7456(c) & (d) (1982), amended by Tax Reform Act of 1986, Pub.L. No. 99-514, § 1556(a), 100 *345 Stat. 2754, codified at 26 U.S.C. § 7443A (Supp.IV 1986).

The language governing the Tax Court's review of a commissioner-Special Trial Judge was originally contained in a rule adopted by the Tax Court in 1973:

(d) Oral Argument and Decision. The Division to which the case is assigned may, upon motion of any party or on its own motion, direct oral argument. The Division inter alia may adopt the commissioner’s report or may modify it or may reject it in whole or in part, or may receive further evidence, or may recommit it with instructions. Due regard shall be given to the circumstance that the commissioner had the opportunity to evaluate the credibility of witnesses; and the findings of fact recommended by the commissioner shall be presumed to be correct.

Tax Court Rule 182(d), 26 U.S.C.App. (1976), 60 T.C. 1057, 1150 (1973) (emphasis added). The Tax Court has since made minor changes in the rule and renumbered it as Rule 183(c), but the last sentence is unaltered. See Tax Court Rule 183(c), 26 U.S.C.App. (Supp.IV 1986). In an explanatory note, the court said that the

[Tax Court] Judge, to whom the case is assigned, may take any action he deems appropriate ..., even with respect to the commissioner’s findings of fact, although they are accorded special weight insofar as those findings are determined by the opportunity to hear and observe the witnesses. In this regard, see Court of Claims Rule 147(b)....

This rule is intended ... to provide procedures more comparable to those which obtain in the Court of Claims.

Tax Court Rule 182(d), 26 U.S.C.App. (1976), 60 T.C. at 1150.

The citation to Court of Claims Rule 147(b) is well founded. Except for the sentence referring to oral argument, Rule 182(d) comes directly from that rule, the major change being that “commissioner” replaces the Court of Claims term “trial judge.” 2

When the Tax Court adopted Rule 182(d) in 1973, the Court of Claims reviewed the credibility determinations of its commissioners under a clearly erroneous standard. Elmers v. United States, 172 Ct.Cl. 226, 232 (1965), explicitly used the term in adopting the commissioner’s findings. In Hebah v. United States, 197 Ct.Cl. 729, 456 F.2d 696, cert. denied, 409 U.S. 870, 93 S.Ct. 199, 34 L.Ed.2d 121 (1972), decided on the eve of the Tax Court’s adoption of Rule 182(d), the Court of Claims adopted a commissioner’s findings, noting that they were “presumed to be correct because of his opportunity to hear the witnesses and to determine the weight to be accorded their testimony.” 456 F.2d at 698. It observed that to overcome the presumption required a “strong affirmative showing.” Judge Davis dissented (with the agreement of two other judges), putting the scope of review expressly into issue. He said that although the court should not overturn a commissioner’s findings without “strong reason,” it was its obligation to overturn the findings “if we are convinced that the preponderance of the evidence goes against the trier’s determination.” Id. at 710.

The Commissioner relies heavily on Montgomery Coca-Cola Bottling Co. v. United States, 222 Ct.Cl. 356, 615 F.2d 1318, 1322 (1980), as did the Tax Court, 3 for the idea that the Tax Court owes little deference to a Trial Judge. Indeed, Montgomery contains generalizations that, loosely read, might be taken to mean that the reviewing court should overturn the *346 initial fact-finder if persuaded of the opposite view by a preponderance:

While we agree the report of a trial judge is entitled to much consideration, that does not impair nor dilute our duty of bearing the ultimate responsibility for determining matters before us. If we are convinced that the preponderance of the evidence goes against the determination of the trial judge, we are obliged to so hold. The presumption of correctness may be applied less stringently to documentary evidence.... [Ojur review of findings based on documentary evidence may be more severe.

... Where we find, contrary to the trial judge’s report, that the greater weight of the evidence discloses that taxpayer’s burden has not been met, this court must substitute its own findings for that of the trial judge.

Id. 615 F.2d at 1322-28 (citations omitted).

When one examines the precise issues in Montgomery, however, it offers far less support for the Commissioner’s view than those generalizations may suggest. The Commissioner asserted that payments of 20 cents per gallon of Coca-Cola syrup, made by several partnerships to an affiliated corporation, were royalties for the use of the corporation’s Coca-Cola bottling franchise. The payments would be personal holding company income if that were the case, but not if they were payments for the use of tangible corporate assets. The trial judge resolved the issue as a matter of the parties’ intent, making findings largely on the basis of the testimony by the taxpayer’s tax accountant. The reviewing court found the focus on the taxpayer’s intent erroneous, id. at 1324, 1328, and said that as a result the trial judge had failed to make “the necessary economic analysis,” id. at 1328, requiring an assessment of “what, economically, a fair payment would be for the franchise,” id. The Montgomery court explicitly quoted the Supreme Court’s decision in United States v. Gypsum Co., 333 U.S. 364, 68 S.Ct. 525, 92 L.Ed. 746 (1948), saying that where the testimony of various documents’ authors

is in conflict with contemporaneous documents we can give it little weight.... Despite the opportunity of the trial court to appraise the credibility of the witnesses, we cannot ... rule otherwise than that Finding 118 is clearly erroneous.

Id. at 396, 68 S.Ct. at 542 (quoted in Montgomery, 615 F.2d at 1328). The Montgomery court’s quotation of language explicitly applying the clearly erroneous test surely indicates no intent to adopt a more freewheeling standard of review.

The court also said that the taxpayer’s “subjective intent testimony ... can only be seriously considered to the extent it is consistent with the objective evidence.” 615 F.2d at 1327. Thus, to the extent that Montgomery is not simply a statement of the substantive legal principles for classification of personal holding company income, it appears to say no more than that reversal of the initial fact-finder is proper if the objective evidence overwhelms the initial fact-finder’s inferences from testimony and demeanor.

Nor do the other cases relied on by the Commissioner do much to assist his case. In two of them the Court of Claims reversed the commissioner on a point of fact, Willett v. United States, 186 Ct.Cl. 775, 406 F.2d 1346 (1969); Miller v. United States, 168 Ct.Cl. 498, 339 F.2d 661 (1964), but in one of those the court found that documentary evidence overwhelmed an inference the commissioner drew solely from a witness’s inability to remember the exact sequence of events occurring 20 years earlier, Willett, 406 F.2d at 1353, and in the other the court found that the commissioner had erroneously preferred the self-serving testimony of the taxpayer as to his intent, as against the flatly contradictory testimony of a neutral witness and undisputed documentary evidence of the taxpayer’s activities, Miller, 339 F.2d at 662-64. Similarly, in Morris v. United States, 171 Ct.Cl. 220 (1965), the Court of Claims reversed the commissioner both because he had applied a “clear and convincing” standard of proof when he should have been satisfied with a “preponderance,” id. at 228, and because “[sjince all the proof was documentary, the trial commissioner was in no special position to judge credibility,” id. *347 at 228 n. 4. Commerce Int’l Co. v. United States, 167 Ct.Cl. 529, 338 F.2d 81, 86 (1964), and Bringwald, Inc. v. United States, 167 Ct.Cl. 341, 334 F.2d 639, 643 (1964), contain ambiguous dicta on the required deference, but have little bite as neither involves reversal on a factual issue at all.

A clearly erroneous standard also follows from the natural reading of Rule 182(d). If a simple “preponderance” of the evidence — half plus a little bit — suffices to overturn the factual findings of a Special Trial Judge, then it is difficult to see what value or force attaches to the presumptive correctness of that judge’s factual determinations, much less the “due regard” owed “to the circumstance that the [Special Trial Judge] had the opportunity to evaluate the credibility of witnesses.” Moreover, reading Rule 182(d) as establishing a clearly erroneous standard gives its “due regard” language the same meaning as the similar phrase has in Fed.R.Civ.P. 52(a). 4 Cf. Tax Court Rule 1 (when no Tax Court rule applies, court shall prescribe procedure, giving particular weight to Federal Rules of Civil Procedure).

The Tax Court is of course free to make its own rules determining the relation between it and its Special Trial Judges. Moreover, we assume that the Tax Court’s construction of its own rules enjoys the deference, on review in this court, enjoyed by an administrative agency interpreting its own regulations. See United States v. Larionoff, 431 U.S. 864, 872-73, 97 S.Ct. 2150, 2155-56, 53 L.Ed.2d 48 (1977) (agency’s interpretation of its own regulation controls if not “plainly inconsistent with the wording”); Udall v. Tollman, 380 U.S. 1, 16, 85 S.Ct. 792, 801, 13 L.Ed.2d 616 (1965) (“When the construction of an administrative regulation rather than the statute is in issue, deference is even more clearly in order.”); San Luis Obispo Mothers for Peace v. United States Nuclear Reg. Commission, 789 F.2d 26, 30 (D.C.Cir.1986) (en banc). But the Tax Court has used language with a reasonably well-established meaning. Indeed, the Commissioner does not cite any prior Tax Court decision under Rule 182(d) manifesting the broad scope of review present here. Thus, until the court adopts new language, it must hew to the meaning of what it has said. Failure to do so rises to a level of inconsistency that even Larionoff does not compel us to endorse.

II.

Having determined that the Tax Court should have reviewed the Special Trial Judge’s factual findings according to a clearly erroneous standard, we now turn to assessing the record. As noted above at page 344, we review the Tax Court under the clearly erroneous standard. Before turning to the specific facts, we take a brief detour to note, but not resolve, a conflict over the exact role of an appellate court where it confronts conflicting decisions by an initial fact-finder and an intermediate court, with the clearly erroneous standard governing both stages of review.

Many of the circuit courts have addressed this matter in reviewing trial court decisions in conflict with fact findings by a special master appointed under Fed.R.Civ. P. 53(a). By virtue of Rule 53(e)(2) (governing trial court use of masters’ findings in non-jury actions) and Rule 52(a) (governing appellate review), “clearly erroneous” controls both stages. A majority approach mimics the Supreme Court in review of a court of appeals judgment, see McAllister v. United States, 348 U.S. 19, 20-21, 75 S.Ct. 6, 7-8, 99 L.Ed. 20 (1954), mod. denied, 348 U.S. 957, 75 S.Ct. 447, 99 L.Ed. 748 (1955), i.e., it simply drops the intermediate forum out. This practice appears to have originated in Learned Hand’s decision in Morris Plan Industrial Bank v. Henderson, 131 F.2d 975 (2d Cir.1942), which reviewed a district court’s review of a referee in bankruptcy:

It is true that logically a distinction can be drawn between holding a referee’s *348 finding [not 5 ] to be “clearly erroneous” and holding a judge’s finding that a referee’s finding is “clearly erroneous” to be “clearly erroneous.” ... We should regret, however, to be compelled now to introduce such refinements into the solution of what is after all only a practical problem.

Id. at 976-77. Compare NLRB v. Universal Camera Corp., 179 F.2d 749, 753 (2d Cir.1950), rev’d sub nom. Universal Camera Corp. v. NLRB, 340 U.S. 474, 492-97, 71 S.Ct. 456, 466-69, 95 L.Ed. 456 (1951), on remand NLRB v. Universal Camera Corp., 190 F.2d 429, 430-31 (2d Cir.1951) (Judge Hand initially favored the appellate court’s completely disregarding a finding by a trial examiner adverse to the agency’s decision, on the ground that there seemed to be no analytically intelligible middle ground between doing so and giving it the deference owed a master’s report, but the Supreme Court adopted the middle ground.).

At least five circuits take the Learned Hand approach, directly reviewing the findings of the initial trier of fact according to the clearly erroneous standard. See Morris, supra; In re Multiponics, Inc., 622 F.2d 709, 722 (5th Cir.1980); O’Rieley v. Endicott-Johnson Corp., 297 F.2d 1, 4-5 (8th Cir.1961); Lines v. Falstaff Brewing Co., 233 F.2d 927, 930 (9th Cir.), cert. denied, 352 U.S. 893, 77 S.Ct. 129, 1 L.Ed.2d 88 (1956); Potucek v. Cordeleria Lourdes, 310 F.2d 527, 530 (10th Cir.1962), cert. denied, 372 U.S. 930, 83 S.Ct. 875, 9 L.Ed.2d 734 (1963). A minority view holds that there is a meaningful difference between direct review and review of review, and that the latter approach, one more deferential to the intermediate court, is the proper one. See Milliken Research Corp. v. Dan River, Inc., 739 F.2d 587, 592-93 (Fed. Cir.1984); United States v. Twin City Power Co., 248 F.2d 108, 112 (4th Cir.1957), cert. denied, 356 U.S. 918, 78 S.Ct. 702, 2 L.Ed.2d 714 (1958).

Our circuit has not resolved the matter, and for this case we find it unnecessary to do so. Even adding some measure of deference to the Tax Court, on the principle that we are deciding whether it was clearly erroneous in finding the Trial Judge’s findings clearly erroneous, we believe that the latter must be reinstated.

At trial, Stone and Rosenbaum admitted the shady origins of the Chromcraft Swiss Funds. They admitted that they kept the Swiss Funds’ existence secret from other Chromcraft corporate officers or directors, and that they included them on neither corporate nor personal tax returns. They testified, however, that in their view the Swiss Funds remained Chromcraft’s property at all times, even though they did not at the times of transfer have specific plans for the use of every dollar. See, e.g., J.A. 184, 186, 199-200. They were able to point to two specific plans, the promotion first of Chromcraft’s rocket launchers, and then later of its Snake Eye Bomb Fin, to NATO countries in Europe. The use of a Swiss account also allowed Chromcraft to collect payments from a Belgian company that was barred from making dollar payments to the United States due to Belgian currency restrictions. J.A. 446. 6 So far as we can determine, neither the Trial Judge nor the Tax Court ever tried to calculate, on their respective readings of the evidence, how much may actually have been expended for the stated promotional purposes. Perhaps both shared the view that the quantity of payments was of limited relevance, since failure to spend them on those plans would not powerfully contradict appellants’ testimony as to their initial purposes (whereas diversion of the funds to non-corporate purposes without repayment of course would have done so). In any event, there is at least some evidence that intervening forces prevented fulfillment of the original pro *349 motional goals: Chromcraft proved unable to sell the Zuni rocket launcher to the United States, J.A. 157, presumably throwing cold water on hopes for NATO sales; and the Snake Eye experienced technical problems in Vietnam, with similar effects on its NATO prospects, J.A. 225, 419-20.

The Special Trial Judge accepted Stone’s and Rosenbaum’s account. He held that the Swiss Funds had remained the property of Chromcraft, and he therefore concluded that the taxpayers were liable for relatively little tax. The Trial Judge recognized that his decision rested in good part on his assessment of Stone’s and Rosenbaum’s testimony, and he explicitly found that testimony credible:

Both of those individuals testified days on end before me, and I found them to be credible witnesses endeavoring to tell the truth regarding numerous and complicated transactions which occurred many years ago. I regard their testimony before me to be more reliable in ascertaining what happened than the testimony in ... [depositions taken six years earlier for the Alsco-Harvard Fraud Litigation ]. Recognizing that both men have pleaded guilty to criminal charges, a fact that must be considered in my evaluation of their credibility, I nevertheless, as stated above, found them to be credible witnesses before me.

J.A. 474. The Trial Judge also noted that in 1973 and 1974 the U.S. government presented Rosenbaum as a witness before a grand jury and in a criminal trial.

Prior to calling Rosenbaum as a witness, representatives of the Department of Justice and the Internal Revenue Service conducted three days of intensive examination of Rosenbaum to assure themselves of being able to vouch for his credibility.

J.A. 352.

The Tax Court did not call any new witnesses, nor did it reexamine any who had appeared before the Trial Judge. It made its decision on the same record as the Trial Judge 7 and explicitly adopted most of his factual findings:

In reaching these holdings, we have to a very large degree, adopted the findings of fact of the Special Trial Judge; to the extent that we disagree with the Special Trial Judge, our disagreement relates to certain inferences he drew and the conclusions which he reached based on those findings.

J.A. 120.

At bottom, the divergence in outcomes appears to arise entirely from the Tax Court’s rejecting the Trial Judge’s determinations as to the credibility of three witnesses as to their understanding of the ownership of the Swiss Funds — Stone and Rosenbaum themselves, J.A. 200, and one Max Vollmer. Because the analysis of the latter’s testimony is somewhat peripheral, we defer it to a special appendix.

Whereas the Trial Judge had found that Chromcraft had a definite corporate purpose when it sent millions to Switzerland, the Tax Court saw Chromcraft as “a victim of [a] massive fraud,” J.A. 287, because

[tjhere were no concrete plans, no specific blueprints for the use of those funds for Chromcraft’s benefit. We are given at most “a pie in the sky by and by,” baked from a recipe prepared by Rosen-baum and Stone when they were unexpectedly presented with a bill they hoped never to receive.

We simply do not believe that if Chromcraft were sold or went bankrupt that the Swiss funds would have been produced by Rosenbaum and Stone for the benefit of the buyers or creditors. They had complete control and dominion over the funds in the same sense as one who embezzles funds, and would have *350 returned them only when and if their defalcations were discovered.

J.A. 287-88.

Examining the record, however, we find little if any objective evidence contradicting the Trial Judge’s decision to credit Stone and Rosenbaum. We start with Stone.

There is no evidence that Stone ever pocketed any moneys from the Swiss Funds, and the Tax Court did not find that he did. He did not have signature power for several of the accounts. He received several loans from various intermediaries who had (at Rosenbaum’s instigation) themselves borrowed from the Swiss Funds, but it was undisputed that in most if not all cases Stone was unaware of the source of those funds. Moreover, it is not disputed that Stone repaid all of these loans in full, and, except for the smallest of them, amounting to less than $7000, he signed interest-bearing notes and repaid with interest. 8 In one case, Stone borrowed approximately $300,000 from domestic sources in order to repay a loan originating from the Swiss Funds that had been channeled through Pacific Enterprises, Inc., J.A. 435, an entity in which he owned no stock and was not an officer or director. J.A. 426. In another case, Stone repaid outstanding loans from Finanz, a German company, by signing over $960,500 in Alsco stock. Unbeknownst to Stone, Finanz too was serving as an intermediary and the money he borrowed came from the Swiss Funds. J.A. 463.

The only even vaguely tenable theory for Stone’s ownership for tax purposes of any portion of the Swiss Funds would be that he engaged in some foresighted joint venture with Rosenbaum so that Rosenbaum would manage the Swiss Funds in a manner that would insulate Stone from any direct dealings with the money. In support of this theory are the Tax Court’s finding that Stone trusted Rosenbaum “like a brother,” J.A. 184 n. 8, and the Trial Judge’s finding that although Stone “never sought to have Rosenbaum disburse any of the monies in the Chromcraft Swiss Fund for the personal benefit of Stone or any of the members of his family, ... Rosenbaum would have acceded to such a request if Stone had made one.” J.A. 430. Extrapolating from this to a meticulously planned conspiracy stretches the evidence farther than it will go. Even if one chose for some reason to ignore the evidence that Stone was not aware that his substantial loans came from the Swiss Funds, there is ample evidence that he felt a powerful obligation to return the money and that he did so. One can, of course, imagine circumstances under which someone would incur debt in the United States in order to replenish Swiss Funds that he believed he owned (either 100 percent or 50 percent), such as clearly higher rates of return on the Swiss Funds. But the Commissioner has not even suggested such a reason, must less offered supporting evidence.

The Commissioner also stresses Stone’s failure to reveal the Swiss Funds to the other shareholder of Chromcraft, whom Stone bought out in March 1965. But given the origins of the funds — for which Stone and Rosenbaum ultimately went to prison — their desire to keep the word within as narrow a circle as possible is by no means inconsistent with viewing the funds as Chromocraft’s. And while Stone’s buying out his fellow shareholder in 1965, still without disclosure, suggests a fraud on that shareholder, it again is quite consistent with his viewing them as corporate.

Thus the objective evidence in no way contradicts Stone’s testimony that he regarded the Swiss Funds as belonging to Chromcraft, and in many respects corroborates it.

The Commissioner’s case against Rosen-baum is more powerful, but in the end insufficient. He had signature power over all the accounts in the Swiss Funds and actively directed their use. Rosenbaum at *351 times commingled the Swiss Funds with moneys he used for investments unrelated to Chromcraft. J.A. 425. He occasionally transferred funds from the Swiss accounts to himself or to his brother, see J.A. 471-73, or used them to make loans to his clients, see J.A. 430-31 (March 24, 1965 loan to Mme. de Ora Trujillo). Rosenbaum asserted that both he and his brother repaid the Swiss Funds with interest, but the Trial Judge found that Rosenbaum had failed to meet his burden of proof regarding the existence of interest payments and therefore disallowed his deduction for them. J.A. 473-74. The most damning item is that during the criminal investigation of his activities he drew unspecified sums from the Swiss Funds to subsidize a business trip to Europe by one Robert Bregman, who had helped transfer funds to Switzerland and was a potential witness against him. J.A. 154.

Despite all this conduct, which to say the least demonstrated total disregard for traditional formalities of bookkeeping, Rosen-baum took other actions which suggest that he did not consider the Swiss Funds to be his to dispose of at will. When he borrowed money from the Swiss Funds for a substantial investment of his own, he promised Stone that if the investment proved unprofitable, he would see that the money was returned to the Swiss Funds with interest, while if it was a success, he would split half the profits with the Swiss Funds. J.A. 427-28. On another occasion, Rosenbaum embezzled $600,000 from an unrelated account in order to place the money in the Swiss Funds. J.A. 461.

Once one rejects the hypothesis that Stone and Rosenbaum together acted as co-owners of the Swiss Funds, this conduct appears impossible to comprehend except as reflecting an assumption by Rosenbaum that those funds belonged to Chromcraft. (No possible owner appears to exist other than Chromcraft, or Stone and Rosenbaum, or Rosenbaum alone.) If Rosenbaum regarded himself as sole owner, why would he feel obligated to indemnify the Funds against any loss when he used them? And why, if Rosenbaum considered the money his own, would he agree to share half the profits from his investments with himself?

There remain the payments to Bregman for his European junket, at a time when Rosenbaum had a strong personal need for Bregman’s cooperation. While these cut against exoneration, they are far from unequivocal. First, there is no indication that the amounts were substantial in relation to the overall sums. Whatever the case for treating these uses as for Rosenbaum’s account, they do not seem large enough to justify regarding him as having acted as owner of the whole $4 million. Second, although Chromcraft obviously could not go to jail, discovery of its financial manipulations carried very severe pecuniary risks, so that the corporation as such stood to gain considerably from increasing Breg-man’s moral and financial indebtedness.

Given the exoneration of Stone, we find that the Tax Court was clearly erroneous in overturning the Trial Judge’s crediting of Rosenbaum (i.e., on our reading of the scope of review, in finding the Trial Judge’s conclusion clearly erroneous). The objective evidence does not materially contradict Rosenbaum’s account, and even tends in some measure to corroborate it.

III.

Because we reverse the Tax Court’s decision on the ownership of the Swiss Funds for tax purposes, there is no need to reach most of the other issues that have been raised by appellants and cross-appellant. There remain a few loose ends. The Commissioner argues that the Tax Court erred in crediting Stone and Rosenbaum with the cash and stock transferred from an account in the Swiss Funds to the United States, a transfer that the Commissioner argues was made to the U.S. Navy in settlement of the Alsco-Harvard Fraud Litigation, The taxpayers argue that they had no interest in the fund, but argue in the alternative that if they did have an interest in the Swiss Funds, the sums transferred to the Treasurer of the United States should be credited against their tax liability. As we reverse the Tax Court’s determination that the taxpayers had an ownership interest in *352 the Swiss Funds, it necessarily follows that they are not entitled to a credit for payments from it.

The Tax Court determined that M. Jeanne Stone was an innocent spouse, but assessed her for taxes as a transferee of property from Andrew Stone. Ms. Stone has contested the assessed transferee liability. Because we reverse the Tax Court, it appears that Andrew Stone’s additional tax liability will fall short of the 25% of reported gross income necessary for M. Jeanne Stone to qualify as an innocent spouse under the controlling version of 26 U.S.C. § 6013(e)(1)(A) (1982). See J.A. 481. This renders her liable for any deficiencies and additions to tax arising from her joint returns with Andrew Stone for the years 1963-67 without regard to any liability as transferee.

Although the Tax Court conducted proceedings to compute the appellant’s taxes under its Rule 156, it did so on the basis of the opinion that we reverse today. We therefore remand the case for computation of tax on the basis of the conclusions of the Trial Judge.

So ordered.

Special Appendix on Testimony of Max Vollmer

Vollmer, a Swiss national, was the (at least titular) head of Establissement Velo International, a company that deposited about half a million dollars into the “Velo Account,” one component of the Swiss Funds. J.A. 223. Vollmer was a signatory on the Velo Account, as were Stone and Rosenbaum. Vollmer testified that he was unaware of any plan to spend the Swiss Funds for development, production or promotion of the Snake Eye. J.A. 223-24. Indeed, Vollmer’s account of his role in the management of Establissement Velo makes him out to be little more than a well-paid clerk. See XV Tr. Document 298 (Direct Examination Questions propounded to Vollmer); XVI Tr. Document 307 (Stone’s Cross-Examination Questions to Vollmer); XVI Tr. Document 310 (Redirect Questions); Petitioner’s Exhibit S-735 (not bound in transcript volumes) (Vollmer’s replies to foregoing); see also XVI Tr. 5855 (S-735 admitted into evidence). Vollmer gave oral replies to written questions administered abroad by an American consular official in Germany. The Trial Judge termed Vollmer’s replies self-serving and untrustworthy because

[a]ll of the questions and cross-questions were furnished to Vollmer in advance, and the impression is that he was giving carefully coached answers in furtherance of his previously declared intention to tell the Government what it wanted to hear, in his own interest.

J.A. 433 (emphasis added). The Tax Court had more faith in Vollmer’s credibility, terming his testimony,

no more self-serving than that of Stone and Rosenbaum.... The vice consul and deposing officer has certified that “on each instance when the witness referred to his notes and memoranda and certain exhibits, it is my opinion that he did so for the purpose of refreshing his memory so as to permit him to testify to matters of his personal knowledge.”

J.A. 224. Because Vollmer was not a live witness, the Special Trial Judge was not uniquely placed to judge his credibility; the Tax Court’s duty to give heightened deference to the Special Trial Judge's judgment as to witnesses’ credibility therefore does not apply. See Morris v. United States, 171 Ct.Cl. 220, 228 n. 4 (1965) (trial judge’s findings regarding demeanor deserve higher deference than findings regarding documentary proof).

That said, the Tax Court appears to have accepted as true testimony that at a minimum can be said to bear earmarks of fiction. As to a $100,000 payment (part of an aggregate of $190,000) transferred from one of the accounts to him and a related corporation, Transmatic, Vollmer was asked, “What did you think the $100,000 was for?” His transcribed answer reads in full:

A. (Consultations with Mr. McNamara [his lawyer, see XVI TR. 5815]) I had no idea.

See XV Tr. Document 307 at 1129 (Stone’s Cross-Examination Questions for Vollmer); *353 Petitioner’s Exhibit S-735 at 34-35 (Vollmer’s replies). See also XVI Tr. 5822 (colloquy among counsel regarding Vollmer’s failure to answer this question). Without an explanation that the Tax Court never provided, we cannot see how the Trial Judge’s heavy discount of Vollmer’s testimony can be viewed as clearly erroneous. In any event, as the Tax Court determined that as a practical matter Rosenbaum, not Vollmer, was “in charge of the Velo bank account,” J.A. 224, it is difficult to see why Vollmer’s professed ignorance as to the purposes of that fund — even if it failed to corroborate Stone’s and Rosenbaum’s account — is sufficiently at odds with Stone’s and Rosenbaum’s account of their intentions to be considered objective evidence discrediting their testimony.

Notes

. The Tax Court wrote:

We have given due regard to the circumstance that the Special Trial Judge had the opportunity to see and evaluate the credibility of the witnesses. Nevertheless, the presumptive correctness of the Special Trial Judge’s report does not impair nor dilute our duty of bearing the ultimate responsibility for determining matters before us. Rule 182(d), Tax Court Rules of Practice and Procedure and note thereto. See 60 T.C. 1150. Montgomery Coca-Cola Bottling Co. v. United States, 222 Ct.Cl. 356, 615 F.2d 1318, 1322 (Ct.Cl.1980).

Rosenbaum v. Commissioner of IRS, T.C.Mem. 1983-113 (Feb. 24, 1983), J.A. 119-20.

. Rule 147(b) of the former Court of Claims read as follows:

(b) Trial Judge’s Report. The court may adopt the trial judge’s report, including conclusions of fact and law, or may modify it, or reject it in whole or in part, or direct the trial judge to receive further evidence, or refer the case back to him with instructions. Due regard shall be given to the circumstance that the trial judge had the opportunity to evaluate the credibility of the witnesses; and the findings of fact made by the trial judge shall be presumed to be correct.

Court of Claims Rule 147(b), 28 U.S.C.App.(1976), rendered nugatory, Claims Court Rules Foreword, 28 U.S.C.App. (1982).

. See note 1 above.

. The parallelism is incomplete, to be sure, as the latter explicitly uses the term "clearly erroneous.”

. The inserted "not” is necessary for the logic of Judge Hand’s statement.

. Although the taxpayers understandably did not press the point, it appears that the diversion may also have served an illegitimate, but very real, corporate purpose by inflating the firm’s apparent expenses for materials used in production and creating an opportunity for fraudulent deductions from Chromcraft’s corporate tax.

. The Civil Division of the Justice Department sought leave to intervene pursuant to 28 U.S.C. § 517 (1982), but the Tax Court denied this request. Instead, the Tax Court accepted the Civil Division’s "Suggestion of Interest” as an amicus brief. Stone and Rosenbaum vigorously argue that this brief, and particularly its allegedly prejudicial attempt to attach preclusive effect to their guilty pleas on charges unrelated to tax fraud, somehow impermissibly harmed their cause. The Tax Court, however, specifically and correctly declined to find any such preclu-sive effect. J.A. 291-92.

. The exception occurred on May 18, 1965 when Rosenbaum wired M. Jeanne Stone $6,689.25 from the Chromcraft account at the Union Bank in Switzerland. Stone did not learn until six years later that the funds transmitted to Paris for his vacationing wife's use came from the Swiss Funds. He repaid the money promptly but, so far as appears, without signing a note or paying interest. See J.A. 187.