The United States appeals the July 13, 1990 judgment of the United States Claims Court awarding to Exxon Corporation (Exxon) a refund of income tax and interest it had paid for tax year 1960. See Exxon Corp. v. United States, 19 Cl.Ct. 755 (1990) (opinion prior to entry of final judgment). Because Chief Judge Smith was permitted, on remand, to revisit and revise the finding of ultimate fact made by the predecessor trial judge, former Chief Judge Kozinski, and because the government has not shown any subsidiary finding on which the new judgment is based to be clearly erroneous, we affirm.

BACKGROUND

The factual background of this lengthy and complex litigation has already been set out in some detail in two Federal Circuit and three Claims Court opinions, ^[1] and we will therefore confine ourselves to a summary of only the facts most pertinent to our disposition of this appeal.

The litigation concerns the income tax consequences to Exxon of Fidel Castro’s takeover of Cuba in 1959 and the Cuban government’s seizure in 1960 of Exxon’s Cuban-based subsidiary, Esso Standard Oil S.A. (known as “Essosa”). ^[2] Specifically, two deductions on Exxon’s 1960 tax return are at stake in this appeal: (1) a $9 million bad debt deduction for a loan Exxon had made to Essosa, and (2) a worthless securities deduction of $2.13 million representing Exxon’s basis in Essosa’s stock. Both parties agree that the allowability of these deductions depends upon the date on which Essosa became insolvent. If, as the government maintains, Essosa became insolvent as a result of a corporate reorganization on June 30, 1960, in which Essosa’s non-Cuban assets were transferred to a newly formed Exxon subsidiary, then both deductions must be disallowed. On the other hand, if, as Exxon contends, the insolvency did not occur until Castro’s government actually seized Essosa’s Cuban assets on the next day, July 1, 1960, then both deductions must be allowed.

This litigation began after the Internal Revenue Service, on audit, disallowed a bad debt deduction on Exxon’s 1960 return for $27.4 million Essosa owed to Esso Export, another Exxon subsidiary, at the time of the Cuban government’s expropriation of Essosa. Exxon paid the tax on this sum under protest, and in 1979 brought suit for a refund. The government then alleged that its earlier allowance of the $9 million and $2.13 million deductions had been erroneous, and argued that they should be off[*876] set against any amount Exxon might be awarded.

The case was tried before Judge Alex Kozinski, then Chief Judge of the United States Claims Court, ^[3] who held that the $27.4 million debt was not worthless because Esso Export had a reasonable chance of recovering it in an action against Exxon, in which it could argue that the June 30 corporate reorganization was a fraudulent conveyance because it left Essosa insolvent. The court therefore rendered a partial judgment dismissing Exxon’s $27.4 million claim. Exxon Corp. v. United States, 7 Cl.Ct. 347, 357 (1985). In his opinion, Chief Judge Kozinski found that Essosa did in fact become insolvent immediately after the June 30 corporate reorganization, id. at 354, though he noted that this finding was in no way necessary to his decision, the dispositive issue being only whether “the question of solvency was sufficiently in doubt so that a fraudulent conveyance action ... would have had a reasonable probability of success.” Id. at 354 n. 8.

Exxon appealed, and we reversed, holding that Exxon was entitled to the $27.4 million bad debt deduction because Esso Export and Exxon were a single entity for tax purposes, and thus Esso Export could have no fraudulent conveyance claim against its parent, Exxon. Exxon Corp. v. United States, 785 F.2d 277, 281 (Fed.Cir. 1986) (“Exxon I”). Judge Kozinski’s findings of fact as to the date of insolvency of Essosa were not addressed in, or in any way material to, our decision in Exxon I.

On remand to the Claims Court, the case was assigned to Chief Judge Kozinski’s successor, Chief Judge Smith. Stating that our opinion in Exxon I had implicitly determined that Essosa’s insolvency was caused by the seizure on July 1 and not the reorganization on June 30, the court ruled that the law of the case doctrine required it to deny the government’s request for offset of the $9 million bad debt and $2.13 million worthless stock deductions. Exxon Corp. v. United States, 12 Cl.Ct. 434, 439-40 (1987). The court accordingly awarded Exxon the tax and interest it had paid on the entire $27.4 million as to which it was entitled to a bad debt deduction, with no offset. Id. at 440.

Upon the government's appeal, we noted that our earlier decision in Exxon I had not addressed, explicitly or implicitly, the question of when Essosa had became insolvent, and held that the Claims Court had erred in ruling that the law of the case required it to allow the $9 million bad debt and $2.13 million worthless stock deductions. Accordingly, we vacated the court’s judgment and remanded for further proceedings on the offset issues. Exxon Corp. v. United States, 840 F.2d 916, 917 (Fed.Cir.1988) (“Exxon II”).

On this second remand, the parties stipulated to having the offset issues decided on the record as it had been developed before Chief Judge Kozinski. Chief Judge Smith concluded that the court’s original finding that Essosa became insolvent on June 30 was based on erroneous assumptions and faulty analysis of the evidence in the record, and found instead that Essosa did not become insolvent until July 1. Exxon Corp. v. United States, 19 Cl.Ct. 755, 764 (1990). The court therefore ruled in favor of Exxon as to both the $9 million bad debt deduction and the $2.13 million worthless stock deduction (though in light of Exxon’s admission that the proper amount of the deduction was only $1.49 million and not $2.13 million, it allowed the government to offset $640,000). Id. at 765. The government appeals.

We have jurisdiction pursuant to 28 U.S.C. § 1295(a)(3) (1988).

DISCUSSION

Two distinct issues are presented in this appeal, which we shall consider in the following order: (1) to what extent Chief Judge Smith had authority to amend the finding of ultimate fact made by his predecessor, Chief Judge Kozinski, and (2)[*877] whether the findings before us in this appeal have been shown to be clearly erroneous.

I

A

The government contends that when the dispositive finding of Chief Judge Kozinski as to the date of Essosa’s insolvency was not disturbed on appeal, it became the law of the case and thus not subject to further revision. This argument misapprehends the scope and meaning of the law of the case doctrine.

In our decision in Jamesbury Corp. v. Litton Indus. Prod., Inc., 839 F.2d 1544, 5 USPQ2d 1779 (Fed.Cir.), cert. denied, 488 U.S. 828, 109 S.Ct. 80, 102 L.Ed.2d 57 (1988) ^[4] , we summarized the doctrine as follows:

The law of the case is a judicially created doctrine, the purposes of which are to prevent the relitigation of issues that have been decided and to ensure that trial courts follow the decisions of appellate courts. The doctrine requires a court to follow the decision on a question made previously during the case____ When a judgment of a trial court has been appealed, the decision of the appellate court determines the law of the case, and the trial court cannot depart from it on remand____ Orderly and efficient case administration suggests that questions once decided not be subject to continued argument, but the court has the power to reconsider its decisions until a judgment is entered.

839 F.2d at 1550, 5 USPQ2d at 1783 (footnotes omitted). ^[5] Law of the case, then, merely requires a trial court to follow the rulings of an appellate court. ^[6] It does not constrain the trial court with respect to issues not actually considered by an appellate court, Holcomb v. United States, 622 F.2d 937, 940 (7th Cir.1980), and thus has long been held not to require the trial court to adhere to its own previous rulings if they have not been adopted, explicitly or implicitly, by the appellate court’s judgment. See Quern v. Jordan, 440 U.S. 332, 347 n. 18, 99 S.Ct. 1139, 1148 n. 18, 59 L.Ed.2d 358 (1979) (“The doctrine of law of the case comes into play only with respect to issues previously determined____ ‘While a mandate is controlling as to matters within its compass, on the remand a lower court is free as to other issues.’ ”) (quoting Sprague v. Ticonic Nat’l Bank, 307 U.S. 161, 168, 59 S.Ct. 777, 780, 83 L.Ed. 1184 (1939)). ^[7]

[*878] Here, though Chief Judge Kozinski’s decision did indeed come before this court, his finding of fact as to the date of insolvency was not addressed in, or in any way relevant to, our decision in Exxon I. Indeed, we explicitly stated in Exxon II that our decision in Exxon I “did not address when Essosa became insolvent.” 840 F.2d at 917. We hold that when a judgment has come before us for review, and certain findings of fact were not examined in, relied on, or otherwise necessary to our decision in that appeal, law of the case does not prevent the trial court on remand from reexamining those findings, With no more deference than if the decision had never been appealed at all. Accordingly, even following our decisions in Exxon I and Exxon II, the Claims Court in this case remained free to amend its earlier finding of fact relating to the date of Essosa’s insolvency.

Nor is the legality of the Claims Court’s revision of its previous findings in any way affected by the fact that the revision was by a different trial judge. A successor judge steps into the shoes of his or her predecessor, and is thus bound by the same rulings and given the same freedom, as the first judge. To the extent that a trial judge can alter a previous ruling, so too can a successor judge:

The first judge always has the power to change a ruling; further reflection may allow a better informed ruling in accordance with the conscience of the court. A fortiori, if the first judge can ... change his ruling, a second judge should have and does have the power to do so as well.

Jamesbury, 839 F.2d at 1551, 5 USPQ2d at 1784 (quoting Corporacion de Mercadeo Agricola v. Mellon Bank Int’l, 608 F.2d 43, 48 (2d Cir.1979)). Here, as Chief Judge Smith explicitly stated in his opinion, the amended finding of ultimate fact was based upon a mathematical analysis of bond data submitted by the parties, and did not rely upon the weighing of conflicting testimony or evaluation of the credibility of witnesses who appeared before Chief Judge Kozinski. Exxon, 19 Cl.Ct. at 761. In this situation, Chief Judge Smith was as free to alter his predecessor’s finding of ultimate fact as Chief Judge Kozinski himself would have been.

B

The government also offers an additional argument as to why Chief Judge Smith should not have disturbed his predecessor’s findings of fact. The government contends that Rule 52 of the Rules of the United States Claims Court (RUSCC), which provides that “[fjindings of fact ... shall not be set aside unless clearly erroneous,” binds Chief Judge Smith in reexamining his predecessor’s findings. This argument misapprehends the scope of RUSCC 52(a). The clearly erroneous standard of RUSCC 52(a), like that of Fed.R.Civ.P. 52(a), applies to appellate review of factual findings. See 9 C. Wright & A. Miller, Federal Practice and Procedure § 2571, at 678 (1971) (“Rule 52 [Fed.R.Civ.P.] requires the trial court to make findings of fact____ The rule also states the effect to be given these findings on review.”) (emphasis added). Though both RUSCC 52(a) and Fed.R. Civ.P. 52(a) are found among those governing trial courts, they recite a standard of review which applies only to appellate review and not to a trial judge’s reconsideration of his own or his predecessor’s findings of fact.

In short, a trial judge like Chief Judge Smith is not required—by law of the case, RUSCC 52, or any other authority cited in this appeal—to give any particular deference to a predecessor judge’s findings of ultimate fact which were not examined in, relied on, or otherwise necessary to the decision in a prior appeal. At least to the extent it was not dependent upon weighing conflicting testimony and evaluation of witness credibility, Chief Judge Smith had authority to amend his predecessor’s finding as to the date of insolvency as he deemed appropriate. It is only this amended finding which we must review in this appeal.

II

We turn then to the question of whether Judge Smith’s amended findings of fact must be reversed. We review find[*879] ings of fact by the Claims Court as to whether they are clearly erroneous. J.R. Cooper v. United States, 827 F.2d 762, 763 (Fed.Cir.1987). The government argues, inter alia, that the judgment before us ignores Chief Judge Kozinski’s weighing of conflicting expert testimony, that it adopts a “risk premium analysis” never presented to Chief Judge Kozinski, and that it ignores what Chief Judge Kozinski considered to be significant “additional risk factors.” Most of these contentions are based on the government’s mistaken assumption that Chief Judge Smith was required to give deference to all of Chief Judge Kozinski’s findings of fact; in essence, the government argues that Chief Judge Smith’s dis-positive finding of fact is clearly erroneous because Chief Judge Kozinski’s finding was not. But Chief Judge Kozinski’s dis-positive finding of fact is not before us for decision, and whether or not it was clearly erroneous is not the issue in this appeal.

In the first place, we note that contrary to the government’s contentions, Chief Judge Smith did not ignore his predecessor’s weighing of testimony. While Chief Judge Kozinski would have been in a better position to evaluate witness credibility, no such weighing of testimony was involved in the decision here, which relied instead solely on mathematical analysis of bond data submitted by the parties. Exxon, 19 Cl.Ct. at 761. Moreover, to the extent the government argues that the judgment on review relies on a method of economic analysis not presented to Chief Judge Ko-zinski, its argument is misplaced. A method of analysis is not comparable to a finding of fact, but derives from law and logic and perhaps, here, economics as well. We thus see little relevance to whether this method of analysis was “before” Chief Judge Kozinski or not. If a trial judge or his successor can change his decisions and findings using a given method of analysis, so too is he free to employ another method of analysis altogether if the one originally used was, on further reflection, determined to be unsatisfactory. A judgment so reached is not reversible unless the new method was impermissible.

Chief Judge Smith explicitly considered the method used by his predecessor, arid gave cogent and thoughtful reasons for ultimately modifying that method of analysis. The government has not persuaded us that Chief Judge Smith’s reasons were unsound or that his method was legally impermissible.

Since Chief Judge Smith’s ultimate finding as to the exact date of Essosa’s insolvency has not been shown to be clearly erroneous, it cannot be overturned. And since, as both parties apparently concede, the question of whether the deductions at issue are allowable depends solely upon that finding, the Claims Court’s legal conclusion that Exxon is entitled to both the bad debt and worthless stock deductions is correct as a matter of law.

CONCLUSION

Chief Judge Smith did not commit reversible error by reexamining and amending the findings of his predecessor. Those revised findings, which are the only ones before us in this appeal, have not been shown to be clearly erroneous. Accordingly, the judgment appealed from is, in all respects,

AFFIRMED.