LARAMORE, Judge.

John Wanamaker Philadelphia, Inc., is a Pennsylvania corporation which operates retail department stores. At the time of the events in issue, it operated a store in Philadelphia, Pennsylvania, and a wholly-owned subsidiary, John Wanamaker New York, Inc., operated a store in New York City. In 1956, the New York subsidiary was merged into the parent company. Two petitions stating substantially identical facts and raising common questions of law have been filed and consolidated. In the interest of simplicity, we shall refer to one plaintiff only and limit discussion to the facts of the Philadelphia store.

As a department store, plaintiff used an accrual method of accounting and a fiscal year ending January 31 for Federal income tax purposes. For the fiscal year ending January 31, 1951, plaintiff elected to value its inventory by the Last In — First Out (LIFO) method. Int.Rev. Code of 1939, § 22(d); 26 U.S.C. § 22(d) (1952 Ed.). As a condition of a LIFO election, the statute required that the opening inventory of the election year and the closing inventory of the preceding year be valued at cost. §§ 22(d)(1)(A), 22(d)(4). This required an adjustment to the fiscal 1950 closing inventory because plaintiff had for many years valued its inventories by the retail method, lower of cost or market, less a reserve for future markdowns. To get the closing inventory of fiscal 1950 up to cost, plaintiff had to increase inventory value by the amount that cost exceeded lower of cost or market, and in addition, plaintiff had to eliminate the reserve for future markdowns. This had the effect of reducing cost of goods sold and thereby increasing taxable income because the Commissioner of Internal Revenue would not allow plaintiff to similarly adjust the opening inventory.

Plaintiff claims a refund of the tax paid for the fiscal year 1950 on the amount of taxable income occasioned by the elimination of the reserve for future markdowns. Both of its arguments are predicated on section 22(d)(3) which provides that a change to LIFO shall accord "with such regulations as the Commissioner * * * may prescribe as necessary in order that the use of such method may clearly reflect income." (Emphasis added.) Plaintiff asserts that "clear reflection of income" is the governing standard we must apply in reviewing the determination of the Commissioner of Internal Revenue. The first argument is that the reserve for future markdowns is illegal under sections 29.22(c)-2 and (c)-8 of Treasury Regulations 111, and therefore must be eliminated from the opening inventory of fiscal 1950, as well as the closing inventory, to clearly reflect income. If the reserve is eliminated from fiscal 1950 altogether, plaintiff is entitled to a refund. The second argument is that the reserve was included in taxable income and exposed to tax in the fiscal years 1916 through 1920 as a result of a decision by the Board of Tax Appeals. Plaintiff says it violates the clear reflection of income standard to expose an item to double taxation.

In opposition, defendant contends that the plaintiff had no right to change to LIFO without first obtaining the Commissioner's consent, and that the Commissioner properly made no adjustment to the opening inventory. Regarding plaintiff's double taxation argument, defendant contends that there is no double tax effect because the reserve in dispute before the Board of Tax Appeals was not the same as the one that is deducted from fiscal 1950 opening inventory.

I. Plaintiff's LIFO Election

Plaintiff must establish that it had an absolute right to elect LIFO to circumvent defendant's argument that the Commissioner can exact a quid pro quo for a change to LIFO. As a general rule, taxpayers cannot unilaterally change from one accounting method to another. Int.Rev. Code of 1939, § 41. The Code gives the Commissioner broad discretion to initiate changes, and, by implication, to police taxpayer-initiated changes. In his regulations, the Commissioner has amplified the statutory rule by providing that a taxpayer cannot change to a new method without permission, and that "[p]ermission * * * will not be granted unless the taxpayer and the Commissioner agree to the terms and conditions under which the change will be effected." Treas. Reg. 111, § 29.41-2. The standard by which the Commissioner's discretion is guided is "clear reflection of income." Defendant refers us to a number of cases which uphold this regulation and allow the Commissioner to exact a quid pro quo. American Can Co. v. Commissioner, 317 F.2d 604 (2d Cir. 1963), cert. denied, 375 U.S. 993, 84 S.Ct. 632, 11 L.Ed.2d 479 (1964); Wright Contracting Co. v. Commissioner, 316 F.2d 249 (5th Cir. 1963); Broida, Stone Thomas, Inc. v. United States, 309 F.2d 486 (4th Cir. 1962); Commissioner v. O. Liquidating Corp., 292 F.2d 225 (3d Cir.), cert. denied, 368 U.S. 898, 82 S.Ct. 177, 7 L.Ed. 2d 94 (1961). To this list we can add our recently decided case, Hackensack Water Co. v. United States, 173 Ct.Cl. ___, 352 F.2d 807 (1965).

The courts have enforced the Commissioner's permission requirement to enable him to protect the fisc most effectively. Any change of accounting method will almost certainly distort net income in the year of change. Although distortion can work either for or against the government, it is a fair assumption that most taxpayer-initiated changes will distort income in the taxpayer's favor, absent some mechanism of control. As a matter of policy, the courts have decided that this area of the revenue laws can be best administered by the Commissioner. The mechanism the Commissioner has adopted is the consent requirement. Thus, he can prevent distortion by withholding his consent until the taxpayer agrees to adjustments that will "prevent the duplication of items of expense or the omission of income with respect to the year of transition * * *." Hackensack Water Co. v. United States, 173 Ct.Cl., at ___, 352 F.2d, at 810. This rule is now codified in section 446(e) of the Internal Revenue Code of 1954.

We are of the opinion that plaintiff was not subject to the requirement of first obtaining the Commissioner's consent with whatever conditions he might impose. We find in Code section 22(d)(1) an absolute right to elect the LIFO method of accounting. The provision states that "[a] taxpayer may use [LIFO]." (Emphasis added.) Throughout the applicable regulations, LIFO is referred to as an "elective method." Section 29.22(d)-1 provides: "This elective inventory method * * * may be adopted by the taxpayer as of the close of any taxable year." Section 29.22(d)-2 requires that the taxpayer must satisfy certain objective conditions precedent. Section 29.22(d)-3 specifies the proper form number for election which is different from the form number used for requesting consent. In addition, this provision empowers the Commissioner to condition approval upon the taxpayer's agreement to use LIFO with respect to goods "other than those specified in the taxpayer's statement of election." That is, of course, a comparatively narrow discretion and far short of the broad power to enforce a condition found in section 29.41-2.

The difficulty arises in the last paragraph of section 29.22(d)-3 and under section 29.22(d)-4, which give the Commissioner power to determine whether to approve an election. The latter provides that a taxpayer may not elect "unless, at the time he files his application for the adoption of such method, he agrees to such adjustments incident to the change to * * * such method * * * in the inventories of prior taxable years or otherwise, as the Commissioner * * * may deem necessary in order that the true income of the taxpayer will be clearly reflected for the years involved." This may look like a broad discretion. It is quite different from the discretion under section 29.41-2, however, because it is expressly limited to one aspect of the election and is governed by a standard, the application of which we can review. It is true that section 29.41-2 is also governed by the clear reflection of income standard, but we feel that the conditions the Commissioner can exact under the notion of "consent" or "permission" there show a much broader discretion and a correspondingly narrower scope of review.

In its brief and oral argument defendant relied entirely on Regulations section 29.41-2. Section 29.41-2 seems to include everything but a LIFO election. The last sentence of the third paragraph instructs us to refer to "[Code] section 22(d) and regulations thereunder with respect to changing to [the] optional method of inventorying goods." There is other evidence that LIFO elections are outside section 29.41-2. Thus, in defining a change in an accounting method, the third paragraph refers to "a change involving the basis of valuation employed in the computation of inventories," and in a parenthetical immediately after, cites all of the inventory regulations except those relating to LIFO. It is interesting to note that this scheme has been preserved in the 1954 Code. In codifying section 29.41-2 of the Regulations under the 1939 Code into section 446(e) of the 1954 Code, Congress inserted the phrase "[e]xcept as otherwise expressly provided in this chapter" to qualify the general rule that taxpayers must secure consent before changing an accounting method. Section 472 of the 1954 Code is the LIFO provision, and as the codification of the old section 22(d) of the 1939 Code, it is elective, subject to regulations "as the Secretary or his delegate may prescribe as necessary in order that the use of such method may clearly reflect income." The regulations under section 472 are almost identical to those in Treasury Regulations 111.

II. The Clear Reflection of Income Standard

The consequence of our concluding that plaintiff was free to elect LIFO in its fiscal year 1951 is that we must review the determination by the Commissioner of Internal Revenue preventing plaintiff from adjusting the opening inventory of 1950. In this task, we are directed generally by section 22(d)(3) of the 1939 Code and specifically by section 29.22(d)-4 of the Regulations to test "adjustments * * * in the inventories of prior taxable years * * * [by a standard that will assure] that the true income of the taxpayer will be clearly reflected for the years involved."

The clear reflection of income test has a long history, dating back to the Revenue Act of 1918, section 212(b), 40 Stat. 1057, 1064-1065. Neither the 1939 Code and its predecessors, nor the regulations, ever defined "clearly reflect income," however. This job has been left to the courts. In determining whether a method of accounting meets the test, courts have looked to "generally accepted accounting principles" in the taxpayer's industry. Where the taxpayer has consistently applied "sound accounting practice," as measured by industry standards, his method has usually met the test, Treas. Reg. 111, § 29.41-2; Sen. Rep. No. 1622, 83d Cong., 2d Sess. 300 (1954); H.Rep. No. 1337, 83d Cong., 2d Sess. 158 (1954), U.S. Code Cong. Admin.News 1954 p. 4025. See Note, Clearly Reflecting Income Under Section 446 of the Internal Revenue Code, 54 Col.L.Rev. 1267 (1954) and cases cited therein. Compare American Automobile Association v. United States, 367 U.S. 687, 81 S.Ct. 1727, 6 L.Ed.2d 110 (1961). Cases in which an accounting method is under attack are very difficult, but it seems that the most difficult cases are those in which the change itself distorts income and requires adjustments to clearly reflect income. The latter are most troublesome because "generally accepted accounting principles" are less likely to provide guidelines for the proper tax accounting of items that have already been deducted and will be deducted again under the changed-to method or of items that have never been deducted and will never be deducted under the new method. These are just representative of the kinds of possible distortions occurring as the result of a change. Unfortunately there is no ready-made test that can be uniformly applied to all "change in accounting method" distortions.

In the present case, we are confronted with a distortion resulting from a change in the method of inventory valuation. This is a problem brought about by the time dimension of accounting. Inventories, of course, continue through time. The closing inventory of one year is the opening inventory of the next. Additions to inventory caused by wholesale purchases and reductions caused by retail sales during the year will determine what is on hand at the close of the year. Any break in the consistent treatment of the value of inventory will distort the cost-of-goods-sold deduction used to compute taxable income. Such a break occurs if for one year the opening inventory is valued by one method and the closing inventory by another. On the facts here, taxable income for plaintiff's fiscal year 1950 has been overstated and distorted because the opening inventory was computed on a different basis from the closing inventory. Plaintiff argues that to meet the "clear reflection of income" test this distortion must be prevented. Because section 22(d)(4) of the 1939 Code unequivocally requires the closing inventory of fiscal 1950 to be valued at cost, plaintiff says we can prevent distortion only by valuing opening inventory on a similar basis. Part of the argument is that the opening, but not the closing, inventory was reduced by a reserve for future markdowns which was an illegal reduction under sections 29.22(c)-2 and (c)-8 of Treasury Regulations 111. The theory is that income cannot be clearly reflected if the accounting method violates the law. The weight of authority is strongly against this argument. There are many cases in which taxpayers have attempted to right a previous accounting error by changing the method. Where the advantage runs solely to the taxpayer via a double deduction or an exclusion of an income item, the courts have said that the prior method, though illegal, was consistent and the changes must be accompanied by an adjustment — presumably to clearly reflect income. Hackensack Water Co. v. United States, supra; Wright Contracting Co. v. Commissioner, supra; Broida, Stone Thomas, Inc. v. United States, supra; Commissioner v. O. Liquidating Corp., supra.

(A) Plaintiff's Double Taxation Argument.

The other part of plaintiff's "clear reflection of income" argument is that we must value opening inventory on a similar basis to closing inventory to prevent double taxation. Again, plaintiff points to its accounting practice of reducing inventory by an arbitrary percentage reserve for future markdowns. It asserts that it has reduced inventories by such a reserve for tax accounting purposes since the introduction of the corporate income tax in 1913. Then plaintiff focuses on the fiscal years 1916-1920 which were before the Board of Tax Appeals in John Wanamaker Philadelphia, Inc. v. Commissioner of Internal Revenue, 22 B.T.A. 487 (1931), aff'd, 62 F.2d 401 (3d Cir. 1932), cert. denied, 289 U.S. 738, 53 S.Ct. 657, 77 L.Ed. 1485 (1933). We are told that the Board of Tax Appeals upheld the government's contention that plaintiffs should not be allowed to reduce opening and closing inventories by arbitrary percentage reserves. The argument goes that this decision had the effect of including in taxable income for the years 1916 through 1920 the reserve reductions to inventories, thereby transforming the arbitrary percentage reserve into a "tax-paid reserve." Although effectively eliminated for years prior to 1921, the old arbitrary percentage reserve, as transformed, was carried forward with additions and subtractions until the year now in issue. Because it is a "tax-paid reserve," it will be taxed again if it is allowed to run through the profit and loss account in fiscal 1950.

To understand plaintiff's argument, it is helpful to review tax accounting practice. Normally, a reserve is created or "funded" by deducting amounts from gross income; these, in turn, are either shown on the assets side of the balance sheet, as reductions to the value of specific assets, or, on the liabilities side, as liabilities reducing assets in general. Those reserves allowable as deductions from income for tax purposes — e.g., bad debt reserves and depreciation reserves, Int.Rev. Code of 1954, §§ 166(c), 167(a) — will result in tax benefits or tax savings for profitable corporations. If for some reason such a reserve is no longer necessary and is therefore eliminated, sound accounting practice requires that the amount of the reserve be run through the profit and loss account. For tax accounting purposes, this has the effect of creating taxable income, the tax on which offsets the prior tax saving. Thus, symmetry is assured by taxing that which escaped taxation previously. An example clarifies: If a taxpayer has been consistently adding to a bad debt reserve by taking year-end deductions, he will have enjoyed tax savings in the amount of the tax he did not have to pay because the income was reduced by the deductions. When he finds that the reserve was not necessary, e.g., if all debtors pay their obligations in full and the taxpayer makes no further loans, he must remove the reserve from his balance sheet and restore it to income. This will be taxable income, and assuming identical tax rates, the applicable tax will exactly offset the prior tax saving.

By referring to a "tax-paid reserve," plaintiff distinguishes its accounting situation from the normal. It argues that its case is unique, because unlike the example above, in which the elimination of the reserve occasions taxable income equal to the prior deduction, it had no prior deduction. Plaintiff concedes that it had deductions prior to 1921, but argues that the effect of the Board of Tax Appeals litigation was to offset those deductions by putting the reserves into taxable income. For fiscal 1921, and subsequent years, however, plaintiff continued to reduce inventories by the reserve "taxed" in the prior litigation. The heart of plaintiff's argument is that to the extent that the reserve, viewed over its entire history, did not result in deductions with consequent tax benefits, it is "tax-paid" — as opposed to "tax-benefitted" — and should not be included in income.

It is easier to understand plaintiff's theory by looking at the figures. Set out below are the amounts deducted from opening and closing inventories of the Philadelphia store for the fiscal years 1916-1920, 1921, and 1950:

PHILADELPHIA STORE

Fiscal year Reserve deducted from inventory ended 1/31 Opening Closing Change

1916 _____________________ $360,407.36 $366,005.00 + $5,597.64 1917 _____________________ 366,005.00 471,710.60 +105,705.60 1918 _____________________ 471,710.60 479,379.62 + 7,669.02 1919 _____________________ 479,379.62 770,583.95 +291,204.33 1920 _____________________ 770,583.95 823,666.35 + 53,082.40 1921 _____________________ 823,666.35 994,048.05 +170,381.70 1950 _____________________ 675,077.72 733,772.27 + 58,695.05

Prior to 1916, plaintiff had built up the reserve to be deducted from 1916 opening inventory to $360,407.36. For the years 1916-1920, the Commissioner of Internal Revenue increased taxable income by adding back the deductions from each year's income. Treating the five years as a unit, this had the effect of increasing plaintiff's income by $823,666.35, which is equal to the 1920 closing inventory. This is the amount of plaintiff's alleged "tax-paid reserve."

The closing inventory of 1920 is, of course, equal to the opening inventory of 1921. Similarly, any reserve carries over. It is axiomatic that this $823,666.35 deduction from 1921 opening inventory yielded no tax benefit. The only tax benefit in 1921 came from the $170,381.70 increase in the reserve. The Internal Revenue Service allowed plaintiff to reduce inventories by this reserve method for all years subsequent to 1920. Tracing the history of these adjustments, we observe that in some years, notably the 20's, reserves rose and with them, tax benefits, but in others, notably the 30's, reserves declined, resulting in taxable income. For the entire period, from opening inventory 1921 to opening inventory 1950, the reserve declined to $675,077.72. The net effect was a $148,588.63 increase in taxable income occurring over the entire period. To be logically consistent, plaintiff points out that it has paid tax twice on this $148,588.63, but it does not ask for relief from that double taxation here. It insists, however, that we should not allow defendant to increase taxable income in the amount of the 1950 opening inventory, which represents the balance of the "tax-paid reserve" or the "already-taxed" reserve.

Central to plaintiff's double taxation theory is the premise that the reserve deducted from opening inventory in 1950 is the same as that in issue in the Board of Tax Appeals litigation. If it is not, it cannot be considered a "tax-paid reserve" because the increased tax resulting from the decision of the Board of Tax Appeals would have been paid on something else and there would be no double taxation. For the purpose of analyzing the applicability of the clear reflection of income standard, we shall assume that plaintiff's premise is correct. In Part III we discuss the infirmities of plaintiff's premise and resolve the controversy over the facts.

(B) Clear Reflection of Income and Double Taxation

Section 29.22(d)-4 of the Regulations provides that adjustments should be made "in the inventories of prior taxable years or otherwise" as are necessary "in order that the true income of the taxpayer will be clearly reflected for the years involved." (Emphasis added.) In this we find a directive to judge whether income is clearly reflected at the time of a change to LIFO not only by reference to the facts in the year of change and the immediately preceding year, but also by reference to facts in all preceding years. If plaintiff did pay tax on the reserve deducted from the opening inventory of its 1921 fiscal year by virtue of the Board of Tax Appeals judgment, and if that same reserve carried over to the opening inventory of 1950, we think that the only way to clearly reflect income "for the years involved" is to exclude from taxable income in the fiscal year 1950 the amount of the reserve to the extent it is "tax-paid." On the assumed facts, the entire reserve should be excluded. This is the only approach that will give effect to the time dimension of inventory accounting discussed above. To determine what the opening inventory of 1950 represents, we must look back over its lifetime; and in looking back, we find that to properly reflect income in fiscal 1950, plaintiff should be able to make the necessary adjustment to take account of the "tax-paid" nature of the reserve.

Our approach is not completely novel. In Singer Sewing Machine Co. v. Commissioner, 5 T.C. 851 (1945), aff'd per curiam, 158 F.2d 982 (3d Cir.) cert. denied, 331 U.S. 837, 67 S.Ct. 1520, 91 L.Ed. 1849 (1947), the Tax Court applied the clear reflection of income standard to facts very similar to those here in issue, and held that the Commissioner could not reduce the opening 1934 inventory of the Singer Sewing Machine Company to the extent that it would cause the double taxation of a "tax-paid" reserve created in 1918 and carried forward in the inventory account. Looking to the facts, the Singer Sewing Machine Company (Singer) bought parts and components from its parent, the Singer Manufacturing Company (Manufacturing) to manufacture products for sale to the public. Manufacturing realized substantial profits on these inter-corporate sales. Singer used the actual price paid to Manufacturing as the inventory value. In 1917, Singer's closing inventory was valued at $12,138,304, which included the $6,106,560.24 profit which Manufacturing had realized on sales to Singer. In all years prior to 1918, Manufacturing and Singer had filed separate Federal income tax returns, so Manufacturing had included the full $6,106,560.24 profit in its taxable income.

In 1918, and in each year through 1933, Manufacturing and Singer filed consolidated tax returns. Revenue Act of 1918, § 240, 40 Stat. 1057, 1081-1082. During the period of consolidation, Singer continued to value its inventories as before. Thus, it bought goods from its parent at prices which included a substantial profit for Manufacturing, and valued its inventory at the actual price paid which included these profits. In reporting income on the consolidated returns, however, Singer treated the inter-corporate profits as a reserve which it deducted from the full price inventory value. For the first consolidated return year 1918, Singer deducted a reserve of $6,106,560.24 from its opening inventory. This was the amount of profit already taxed in 1917 and prior years. Without more, this reserve deduction would have increased income in 1918. In fact it did not, because Singer reduced 1918 closing inventory by a reserve for inter-corporate profits of $8,245,474.67 (the $6,106,560.24 plus Manufacturing's profit during 1918) which decreased income more than the increase caused by the adjustment to opening inventory. Through 1933, the inter-corporate profits were deducted as reserves from both opening and closing inventories.

After 1933, Manufacturing and Singer were not allowed to file consolidated returns. Revenue Act of 1934, § 141, 48 Stat. 680, 720-722. Accordingly, in 1934 they filed separate returns. The value of Singer's 1933 closing inventory was $18,422,333.36. The reserve for inter-corporate profits was $8,431,575.92. The Commissioner of Internal Revenue had issued regulations which required Singer to reduce its opening 1934 inventory (but not closing 1934 inventory) by the entire $8,431,575.92 reserve. Treas. Reg. 86, art. 113(a) (11)-1. This had the effect of increasing 1934 income by the full amount of the reserve. The supposed purpose of this adjustment was to prevent the inter-corporate profits from forever escaping taxation. The Tax Court, applying the clear reflection of income standard in section 141(b) of the Revenue Act of 1932, 47 Stat. 169, 213, reviewed all the years from 1918 through 1933 to determine the net effect of Singer's treatment of inter-corporate profits. It found that during the entire period in which consolidated returns had been filed, only $2,325,015.68 of the $8,431,575.92 reserve had escaped tax. The other $6,106,560.24 had already been included in income by Manufacturing in 1917 and prior years. The court held that this amount — this "tax-paid" portion of the reserve — should not be taxed a second time. To clearly reflect income, it reduced Singer's opening 1934 inventory only by the $2,325,015.68 that had not been previously taxed.

We have labored the facts in Singer because they are so close to the present case. The reserve which the Wanamaker Philadelphia store deducted from its opening fiscal 1921 inventory became a "tax-paid" reserve by virtue of the litigation in the Board of Tax Appeals. The reserve which the Singer Sewing Machine Company deducted from its opening 1918 inventory was a "tax-paid" reserve or "already-taxed" reserve by virtue of the prior taxation of its parent's profits. When the Commissioner of Internal Revenue required Singer to reduce its opening 1934 inventory by the reserve, the Tax Court held that to the extent this would tax again the amount already taxed it could not clearly reflect income. The Commissioner has required plaintiff in this case to reduce its opening 1950 inventory by the reserve, and we can say now that this does not clearly reflect income because it exposes to tax the balance of the amount already taxed.

Defendant would have us distinguish Singer on its facts. The theory is that the Tax Court was really trying to prevent the government from taxing Singer on a profit properly realized by Manufacturing. That was decidedly not the basis of the Tax Court's holding. There is no question the holding is grounded on the application of the clear reflection of income standard. Singer Sewing Machine Co., supra, 331 U.S. at 855, 67 S.Ct. 1520. Defendant also argues that plaintiff seeks to circumvent the statute of limitations and to ignore the rule that tax liabilities are determined on an annual basis. See Burnet v. Sanford Brooks Co., 282 U.S. 359, 51 S.Ct. 150, 75 L.Ed. 383 (1931). And in oral argument, defendant stressed that plaintiff is really asking for mitigation for which it does not qualify. See Int.Rev. Code of 1939, § 3801; Int.Rev. Code of 1954, §§ 1311-1315; H.T. Hackney Co. v. United States, 111 Ct.Cl. 664, 78 F. Supp. 101 (1948); Gooch Milling Elevator Co. v. United States, 111 Ct.Cl. 576, 78 F. Supp. 94 (1948). See also Int.Rev. Code of 1954, § 481. These appear to be red herrings, out of the context of the proper meaning and application of "clearly reflect income." In making an historical survey of the Wanamaker inventories to determine whether income is clearly reflected in the fiscal year 1950, we do not think we are doing any violence to the tax accounting concepts defendant relies upon. We are not "opening up" prior years or spreading income or deduction items about. We merely refer to the prior years for certain objective facts, a practice which is quite common in tax law and particularly appropriate here because inventories are a part of a chain, the origins of which can only be learned by inquiry into the past.

III. The Fact Issue

Until now, we have assumed, as contended by plaintiff, that since 1913 the deductions from inventories were "illegal arbitrary percentage reserves." Treas. Reg. 111, §§ 29.22(c)-2 and (c)-8. The defendant sharply contests this contention, arguing that the reductions to inventory considered by the Board of Tax Appeals were not entirely the same and that plaintiff is collaterally estopped from asserting the contrary now. The Trial Commissioner set the stage for this argument because he made findings of fact that the reserves deducted in 1916-1920 were not composed exclusively of "illegal arbitrary percentage" mark-downs. Apparently, during the pre-trial and trial proceedings, and indeed through the proposed findings of fact stage, the parties were in substantial agreement that the reserve had always been composed exclusively of "illegal arbitrary percentages." Plaintiff, understandably, would like us to bind defendant by its earlier concession inasmuch as there is no double taxation unless the item in 1950 is the same as that in 1916-1920. We are of the opinion, however, that it is neither desirable nor necessary to limit defendant to its original theory of the case. We substantially agree with our commissioner's, and defendant's, present view of the facts, but conclude that what the plaintiff was doing to reduce inventories in the years 1916 through 1920 was functionally the same as what it did through 1950. This is established by the preponderance of plaintiff's evidence — even though plaintiff did not choose to argue its case this way.

The Board of Tax Appeals sustained the decision of the Commissioner of Internal Revenue disallowing the 1916-1920 deductions from inventory on the ground that the Wanamaker companies had failed to prove that the deductions were other than "illegal arbitrary percentages." 22 B.T.A., at 500. The Board said, however, that the method as explained by plaintiff's witnesses was proper (at least in part) if provable — i.e., the reserves were not entirely "illegal arbitrary percentages." 22 B.T.A., at 501. Because the companies had lost the necessary accounting worksheets, allegedly due to the government's negligence, the only proof they could offer was through the testimony of their executives. There is no doubt that these witnesses testified that the reserves were other than arbitrary percentages, non-deductible then as now.

We have reviewed all the testimony before the Board of Tax Appeals to ascertain the basis of its findings of fact and holding, and get the following picture: On the last day of each fiscal year, Wanamaker sales clerks counted all items in their departments and entered on tally sheets both the number of items on hand and their retail prices. Store buyers reviewed the sheets to enter the cost. In a few instances, the buyers would enter a figure lower than cost, reflecting personal knowledge of a decline in value. For most items, however, only the merchandise managers made the necessary adjustments to get the inventory to a "true" lower of cost or market value. This process took place over a 2-week period, the manager responsible for each store (in Philadelphia and New York City) spending about 100 hours. Because the inventory consisted of hundreds of thousands of items, the review was necessarily selective. Typical considerations for reducing the stated cost of the item to arrive at market value were the age and condition of the merchandise, its vulnerability to fashion changes, and the sizes and colors available, to name a few. Then there were deductions to "cost" as such, as contrasted with deductions from cost to get to market. The one most talked about was the adjustment to the cost of foreign source merchandise to reflect changes in foreign currency values. Also, invoices were occasionally lost or inaccurate, necessitating cost adjustments. The total of these adjustments and an additional adjustment to reflect a future markdown for the forthcoming February sale was equal to a reserve. In addition, the Wanamaker companies set up a reserve of a flat 10 percent of the inventory as reduced by all the adjustments. That 10 percent reserve further reduced inventories, but brought no tax deduction and was accordingly not in issue. To sum up: The uncontested testimony of the Wanamaker witnesses in the earlier litigation shows that the item today called the "illegal arbitrary percentage reserve" was for 1916 through 1920 apparently a composite of reductions used to get inventory to the "true" lower of cost or market. Considering the tremendous number of items adjusted, probably most of the reductions were arbitrary percentages. However, only the future markdown for February sales adjustment was clearly "illegal" under the predecessor to Treasury Regulations 111, §§ 29.22(c)-2 and (c)-8. Because the records were not available, the witnesses were not able to identify what portion of the "reserve" was composed of this "illegal" adjustment.

In the trial in the present case, the parties agreed that an assistant treasurer of Wanamaker would be the only witness to develop the facts relating to Wanamaker's inventory accounting practice. This witness had been employed by the plaintiff since 1938, and consequently had no personal knowledge of the accounting practice in the years before the Board of Tax Appeals. His testimony covered the earlier years, however, and he could not be shaken from his position that for 1916 through 1920, and indeed for all subsequent years, the inventory accounting method was essentially unchanged. With respect to the testimony in the earlier case, he said on direct examination that the so-called "illegal arbitrary percentage reserve" was computed by totalling the arbitrary percentage deductions for all departments. He explained the non-deductible flat 10 percent reserve (which was not before the Board) as a general reserve computed as a percentage of the entire inventory. The witness obviously appreciated that his interpretation of the facts was inconsistent with the testimony of the witnesses before the Board of Tax Appeals. Thus, on cross-examination, he conceded Wanamaker contended the reserve was not an "illegal arbitrary percentage reserve" in 1916-1920, but he explained that Wanamaker understood the judgment in that case as meaning that the reserve was an "illegal arbitrary percentage reserve." The witness testified further that the reason Wanamaker continued to use the "arbitrary percentage reserve" method after it "knew" it was illegal (certiorari was denied in 1933) was that it considered the reserve to be "tax-paid."

We have summarized the testimony in the Board of Tax Appeals trial and in the trial in this court to show that there is an irreconcilable conflict. In the earlier case, Wanamaker wanted a tax deduction which it could have only if it could prove that the reserve was not the composite of arbitary percentage writedowns, prohibited by the regulations. In the present case, there is no question that the reserve deducted from opening 1950 inventory was an "illegal arbitary percentage reserve," so the plaintiff has tried very hard to convince us that the reserve in 1916-1920 was the same thing. Both parties argue that the decision of the Board of Tax Appeals has a collateral estoppel effect. Plaintiff says that we should look to the judgment only, which had the effect of disallowing deductions in the amount of the reserve; defendant says we should look to that part of the decision which suggests that the method of accounting as described by plaintiff's witnesses was proper and by implication not an "illegal arbitrary percentage reserve" method. 22 B.T.A., at 500.

In our view, it is not necessary to rule on the question of who is estopped. We think the testimony in the trial in this court, coupled with the testimony before the Board of Tax Appeals and its findings of fact and opinion, establish that the Wanamaker accounting practice during the fiscal years 1916 through 1920 was different from that in 1950. No one will ever know whether the reserve in the earlier years was entirely an "illegal arbitrary percentage reserve" as the Internal Revenue Service then found it convenient to contend, and as plaintiff asks us to decide now. We assume that it was not, or that only part of it was. We focus on the purpose of the reserve method used in 1916 through 1920, and conclude that plaintiff was using a modified "cost method, lower of cost or market," inventory valuation system, which was designed to produce a "true" lower of cost or market inventory value. This was a conservative bookkeeping practice which "squeezed the water" out of inventories and produced a more realistic balance sheet.

At some time between 1921 and 1950, the reserve definitely became entirely an arbitrary percentage reserve, clearly illegal under the regulations. We do not know the circumstances of this transformation. It appears that it occurred in the fiscal year 1935 when plaintiff discontinued using the "cost method, lower of cost or market" and began using the "retail method, lower of cost or market." The retail method was selected because its use eliminated the tremendous job of adjusting the cost of each item of merchandise. In computing a retail method inventory value, an accountant need only know the retail price which he simply reduces by the average percentage mark-on. In shifting to this method, plaintiff continued to reduce the inventory value by the "tax-paid" reserve carried over from earlier years. In 1935 and thereafter, the reserve was computed by taking a percentage of the retail method inventory value, the percentage varying with the age of the merchandise. For tax accounting purposes, this was clearly illegal under the regulations. Again, we focus on the purpose of the reserve method, but this time on the method used after 1934, and we conclude that, just as we said about the method used by plaintiff in 1916 through 1920, plaintiff was using the reserve as a modification to the "retail method, lower of cost or market" inventory valuation system, to produce a "true" lower of cost or market inventory value.

IV. Conclusion

Because we conclude that the function of the reserve before 1934 was identical to the function after, and that the "old" reserve was merged into the "new," we are of the opinion that plaintiff can recover even though we cannot say the reserve has always been solely the composite of "illegal arbitrary percentage" deductions. In so holding, we do not intend to cloak the "illegal arbitrary percentage reserve" method with legality. There is no question that sections 29.22(c)-2 and (c)-8 of the Treasury Regulations 111 forbid this method. Illegal as it may have been, however, plaintiff consistently followed this practice until 1950 without any objection by the Internal Revenue Service, and in so doing, carried forward the old "tax-paid reserve. We agree with the reasoning in Singer Sewing Machine Co., supra, as it relates to this case, and hold that to clearly reflect income at the time of the change to LIFO, the Commissioner of Internal Revenue cannot reduce the plaintiff's opening 1950 inventory by the amount of the "illegal arbitrary percentage reserve." This will give plaintiff a substantial recovery, but one to which it is entitled because it has already paid its tax.

Judgment shall be entered for plaintiff John Wanamaker Philadelphia, Inc. in No. 305-61, and for John Wanamaker Philadelphia, Inc. (Successor by merger to John Wanamaker New York, Inc.) in No. 70-63, with the amount of the refund, including interest, to be determined pursuant to Rule 47(c)(2).