dissenting.
This case involves construction of a rather opaquely worded provision of the Internal Revenue Code of 1939, § 122. But the problem to which this section is directed, its objectives and the general plan by which they are pursued, ought not to elude clarity.
Our system of income taxation operates on an annual basis. Each taxpayer is required to determine, on the basis of knowledge available to him at the end of the taxable year, the amount of income or loss for that year. In its original strict form, this system did not permit readjustment of the annual income figure to reflect unanticipated events occurring in subsequent years— for instance, repayment in a later year of money received from sale of goods or services and reported as income in the earlier year — even though logically and practically these facts operated to reduce the income as originally reported. More important, the system required that a taxpayer with profit in one year and an equal loss *244in the next year pay taxes in the year of profit without regard to the loss in the next year, even though from a business and human point of view the taxpayer, over the. broader period of two years, had no income.
Here we are dealing with certain ameliorations of the unduly drastic consequences of such a system in its rigid form. Primarily we have to consider § 122 of the Internal Revenue Code, which, to a limited degree, permits reflection of the fact that income in one year may, in the not uncommon fluctuation of business affairs, have been offset by losses incurred in subsequent or preceding years. At the time here relevant, § 122 provided that if a “net operating loss” in business operations occurred in one year, and net income had been or was later received in any of the two preceding or succeeding years, this net income could be cancelled against the loss by “carry-back” or “carry-over” of the loss to the year in which income was received. If the loss were carried back, reducing the income already reported, taxes already paid on the amount of income cancelled were to be refunded.
Stripped of details, the scheme appears simple. However, with a view to dealing comprehensively with the multifarious manifestations of business activities, the scheme as embodied in intricate statutory form raises difficulties. They are accentuated in this case because of the relevance of other peculiarities of tax accounting.
Thus, the first question raised in this case, one already dealt with in United States v. Olympic Radio & Television, Inc., ante, p. 232, brings into focus the differences between two tax accounting systems, one in which the year’s transactions are recorded on a cash receipts and disbursements basis, and the other utilizing the accrual system. We agree that this taxpayer, whose taxes for all other purposes are calculated on the accrual system, must determine the amount of the loss which may be carried forward or back on the same basis.
*245The other question, however, brings into play the more complex provisions of § 122 as well as a second statutory modification of the strict annual concept. This modification grows out of the “renegotiation” system devised by Congress to deal with potentialities of unconscionable profits to war contractors providing supplies to the Government. Under it, the contract price originally agreed upon between the contractor and the Government is subject to “renegotiation” at a later date to determine whether what originally had been thought to have been a fair price in fact proved overly generous. If so, the Government was entitled to a reduction in the price or, if payment had already been made, a refund of the disallowed profit. Naturally enough, the taxpayer was permitted to reflect the fact that he had been required to repay amounts on the basis of which he originally paid income taxes by reopening the earlier tax return, reducing the income reported and the tax due. 26 U. S. C. §3806 (b).
We are concerned here with a taxpayer on the accrual basis which in 1946 suffered a net operating loss ($164,-326.38), but in 1944 and 1945 had a considerable net income ($827,852.991 and $1,215,320.25 respectively). *246In 1944 and 1945 petitioner was therefore required to pay both corporate income and excess profits taxes. Much of the income accrued by petitioner in the year 1944 was derived from war contracts. Renegotiation in subsequent years reduced the amounts actually received or retained by petitioner below the figures reported for 1944 by $397,970.00. Petitioner, therefore, received a refund of excess profits taxes reflecting this fact.
On the basis of petitioner’s operating loss in 1946, it is entitled to carry back this loss under § 122. Where, as here, the taxpayer has had net income in both of the two years preceding the loss year, that section directs that the carry-back be first applied to the earlier taxable year, 1944, and the remainder, if any, be applied to reduce income for 1945. It is not in dispute that petitioner could *247carry back to 1944 the full amount of the net operating loss which incurred in 1946, and petitioner’s income and excess profits taxes for that year have been reduced to the full extent of this loss. Does anything remain to be applied in reduction of 1945 income?
Congress has provided a detailed formula for determining how much loss remains for application in 1945. The difficulty arises in applying the formula, because Congress has not made explicit, and so left in doubt, what set of figures may fairly be used in translating the generality of the formula into amounts. The difficulty is enhanced because the relevance of some of the factors used in the formula to any discernible congressional purpose is unclear. Logically one would expect that the statutory formula would be designed merely to permit application *248to 1945 of so much of the 1946 loss as was not used to cancel 1944 taxes. But Congress, for its own good reasons, felt that certain other factors should be reflected in this calculation. One factor, which it is not for us to explain, was the amount of the taxpayer’s excess profits tax for 1944.
To vivify this problem, one must reduce to technical concreteness the statutory formula. It states, insofar as here relevant, that the amount of carry-back left for application in 1945 is the amount of the 1946 loss, less the figure arrived at by subtracting from “the net income for the second preceding taxable year [1944],” “the amount of tax imposed by Subchapter E of Chapter 2 paid or accrued within the taxable year [1944].” The latter is a dry statutory description of the excess profits tax.
Thus in addition to petitioner’s 1944 net income, the formula makes the 1944 excess profits tax figure crucial. The question here is: what 1944 excess profits tax figure? The amount of the tax due on the basis of the taxpayer’s knowledge at the close of the 1944 taxable year, that is, what its truthful balance sheet for that year indicated to be the tax ($625,561.59)? Or the amount of the tax which petitioner eventually and definitively had to pay after subsequent events had resulted in a downward revision of the originally reported 1944 tax ($318,577.67 tax reduction due to renegotiation, plus approximately $150,000.00 due to the carry-back of the 1946 loss)?
If it is the former figure, petitioner’s excess profits tax reported in 1944 ($625,561.59) was larger (because it did not take into account the then unknown reduction due to renegotiation and carry-back) than the figure for 1944 net income ($584,866.81), which, for reasons later to be explained, all parties concede should be the figure used in the formula, one reflecting the fact of later renegotiation, but not reflecting the 1946 loss carry-back. Applying the formula on the basis of the larger excess profits *249figure, there is nothing to subtract from the 1946 loss, and the full amount of this loss is therefore available to offset 1945 net income and bring about a further refund of taxes for that year. But the full amount of the 1946 loss has already been applied in 1944 to offset the 1944 income and to bring about a refund for that year. Thus the Court’s decision permits the loss in 1946 to offset twice as much income in 1944 and 1945.
If, on the other hand, the 1944 excess profits tax figure is adjusted to reflect the reduction in the tax occasioned by renegotiation and the 1946 carry-back to 1944, the formula will not permit such double use of the 1946 loss; the difference between the 1944 net income figure ($584,866.81) and the adjusted 1944 excess profits tax ($280,540.33) is greater than the amount of the 1946 loss ($164,326.38).
Either of these positions can be supported by arguments based solely upon the literal language of the statute. Here we are not compelled in our choice by austere regard for what Congress has written, undistorted or unmitigated by judicial rewriting, no matter what the consequences in a specific case. Where the taxing measure is clear, of course, there is no place for loose conceptions about the “equity of the statute.” Revenue laws are notoriously not expressions of an ordered system of reason and fairness. There has probably never been a revenue statute which, by design or oversight, has not favored some groups and laid the basis for a claim of unfairness to others similarly situated. But while one should sail close to the shore of literalness in dealing with the technical problems which are the subject matter of revenue laws, literalness of meaning affixed merely to a particular word or phrase may itself distort what the provision as an entirety and in context conveys and therefore commands. And where ambiguous language is used, the mode of construction applied in Olympic Radio, supra, *250should not be ignored — that deductions and credits are matters of legislative grace and the taxpayer must bring himself clearly within the relief he claims.
Nor does the decision in Olympic Radio shed any light on our present problem. In that case we merely decided that an accrual taxpayer must be consistent in his choice of tax accounting systems — that in calculating his loss for a given year he must use figures relevant to that year’s operations, accrued figures, rather than figures based on cash payments related to transactions in a different year. We had no occasion to determine in that case whether the accrued figure to be used was to be the figure originally reported or the figure reflecting later adjustments to the original figure. That is the issue here.
The crucial phrases in determining whether the original or the adjusted figure for 1944 excess profits tax is to be used are: (1) “tax imposed by Subchapter E of Chapter 2,” and (2) “paid or accrued within the taxable year.” We agree that the first phrase serves merely to identify the nature of the tax referred to and vouchsafes nothing on the question in issue. We need not add to what the Court has said on this. We do not agree, however, that the second phrase compels the reading which the Court gives it.
For purposes of payment of current annual taxes, the phrase “accrued within the taxable year” has come to mean the figure arrived at by taking into account only knowledge available at the end of the taxable year. This reflects the fact that our tax system is operated on an annual basis and that it would be difficult to permit continual reopening and readjustment of old returns in light of later developments. See Burnet v. Sanford & Brooks Co., 282 U. S. 359; Security Flour Mills Co. v. Commissioner, 321 U. S. 281, 286; Dixie Pine Products Co. v. Commissioner, 320 U. S. 516, 519. Were we presented with a question whether the taxpayer owed the Govern*251ment interest on a deficiency, existing at the time his return was filed, in payment of “accrued” taxes, we would agree that the tax “accrued” was the tax calculated on the basis of the situation at the end of the taxable year without regard to any later adjustment in the amount of tax due which eliminated the deficiency. Cf. Manning v. Seeley Tube & Box Co., 338 U. S. 561; United States v. Koppers Co., 348 U. S. 254.
But if “accrued” has this meaning generally in our taxing system, it has acquired this sense not because it inevitably, lexicographically speaking, has this meaning, but because of the inferences which have grown up about it through use in the context of annual payment of taxes. In short, usage, the ultimate glossator, has made it a term of art in this context.
In the present case, we deal with sections of the Code which express exactly the opposite philosophy from that which gave rise to this use of “accrued” as a technical term looking only to events occurring within a single year. We deal with sections which direct re-examination of returns for past years in an effort to ameliorate the shortsightedness of the annual system which fostered a restrictive, closed meaning of “accrued.” The very purpose and direction of these sections require adjustments to earlier returns on the .basis of subsequent facts. Surely in this context there is no rational reason for refusing to recognize the state of affairs as unfolded in the years which § 122 directs you to re-examine — the current year and two preceding years. Where the subsequent events are recognized by the Code as proper occasions for adjusting old returns, the arguments of administrative convenience which underlie closing the tax affairs of the year within the taxing year are empty because they have nothing to which they can apply.
Section 122 was designed to relieve the taxpayer from an unrealistic concept which taxed income which really *252was not there. Surely we should not conclude that Congress intended to go to the other extreme and refund taxes on the basis of losses which are really not there. There is confirmation within § 122 itself which should preclude such result. It lies in subsection 122 (b)(1)(B),2 which alone justifies the assumption, made by the Court and both parties, that the figure for 1944 net income to be used in the formula is the lower figure reflecting the adjustment for renegotiation ($584,866.81) rather than the one originally reported ($827,852.99). All the argument used to demonstrate that the 1944 excess profits figure should be the unadjusted figure proves with equal force that the 1944 net income figure should be the unadjusted figure. For “net income” for any given year is, generally speaking, arrived at by an accrual taxpayer by determining accrued gross income and accrued deductions for that year. §§ 21 (a), 22, 23, 41, 42 (a), 43, 48 (c). Inherent, therefore, in § 122’s phrase “net income” is the same concept of “accrued” as is explicitly used in that section’s reference to the excess profits tax. If “accrued” does not permit taking into account later adjustments to the tax, it does not permit reflection of adjustments to net income.
But subsection 122 (b)(1)(B) shows clearly that Congress assumed that the formula we have been discussing would reflect subsequent adjustments to 1944 net income. That subsection states in effect that the figure for net income of 1944 used in the formula is not to reflect the fact that the 1946 net operating loss has already been carried back and applied to reduce 1944 net income. If Congress thought it necessary specifically to direct that a certain adjustment to 1944 net income arising from facts developed in later years should not be made, this can only be *253because Congress assumed that in applying the formula the figures used would reflect such adjustments. We recognize this assumption by taking as the 1944 net income figure the originally reported figure less the amount by which it was reduced as a consequence of renegotiation. There is no basis for differentiation, in recognizing this assumption, between net income and excess profits tax.
I would affirm.
For convenience of reference and because in most instances it makes no difference in the ultimate result, the figures here used for petitioner’s 1944 net income and excess profits tax are those which have been used throughout this litigation in the briefs, arguments and opinions. The fact of the matter is that there are in the record three conflicting sets of figures which purport to represent petitioner’s net income and excess profits tax in 1944. In one contingency, to be noted below, it becomes important to distinguish between these figures.
Petitioner filed two returns for 1944, an original return and an amended return. The Commissioner objected to several deductions which were claimed on the amended return and also permitted several deductions not claimed by petitioner on the amended return, thus arriving at a corrected return for 1944 substantially different from either the original or amended return. Throughout the litigation the figure cited for petitioner’s 1944 net income, unadjusted to *246reflect subsequent events, has been $827,852.99. This is in fact the figure which petitioner reported on its original return as its excess profits net income. The net income figure on the original return was $817,680.90. Petitioner’s amended return showed a net income for 1944 of $747,236.60, while correction of petitioner’s amended return to reflect errors found by the Commissioner resulted in a figure of $982,836.81.
Similarly the accrued excess profits tax for 1944, unadjusted to reflect subsequent events, has been given as $625,561.59. This is the raw figure which petitioner’s original return showed as his tax before certain credits and adjustments which he claimed. That return showed as the excess profits tax due for 1944, $605,561.59. The amended return showed $549,206.15 as the tax due. Becausé of the fact that at the time the Commissioner corrected the errors in petitioner’s amended return he was aware of and took cognizance of the later renegotiation and 1946 carry-back, he did not arrive at what would have been the correct figure for the tax which petitioner should have originally reported as due for 1944 on the basis of facts known at the time the 1944 return was filed. The figure would, however, approximate $770,000.00.
When, however, the figures for either the 1944 net income adjusted to reflect renegotiation, or the 1944 excess profits tax after adjustment to reflect renegotiation and loss carry-back, have been given, they are the figures arrived at by the Commissioner ($584,866.81 and. *247$280,540.33 respectively). Since these figures are based not only upon a different return by petitioner, but reflect substantial corrections by the Commissioner of the taxpayer’s errors, they are not comparable to the figures which are quoted for unadjusted 1944 net income or excess profits tax. Fortunately for most purposes use of these figures does not affect the conclusion reached on the only question involved here — whether any loss is left to carry back to 1945.
There is one point at which use of comparable figures does become important. If the figures which have been in general use were comparable, then in our disagreement it would be necessary only to conclude that the unadjusted excess profits tax figure for 1944 ($625,561.59) should be reduced by the amount of tax refunded because of the renegotiation ($318,577.67). Upon application of the formula, subtraction of the resultant excess profits tax figure ($306,983.92) from 1944 net income adjusted to reflect renegotiation ($584,866.81) would give a figure in excess of the 1946 loss carry-back and thus preclude any carry-back to 1945. But if comparable figures are used in applying the formula — it does not matter which of the three sets of figures is used — the reduction in 1944 excess profits tax due merely to renegotiation does not preclude a carry-back to 1945 and thus it is necessary to conclude that the excess profits tax figure for 1944 must also be reduced to reflect the refund resulting from the carry-back to 1944 (approximately $150,000). So reduced, there is no carry-back to 1945.
This subsection provides that in applying the formula, the figure for net income of 1944 is to be “computed . . . (B) by determining the net operating loss deduction for such second preceding taxable year [1944] without regard to such [the 1946] net operating loss.”