Mr. Justice Clark
delivered the opinion of the Court.
These consolidated cases involve the depreciation allowance for automobiles used in rental and allied service, as claimed under § 23 (1) of the Internal Revenue Code of 1939, which permits the deduction for income tax purposes of a “reasonable allowance for the exhaustion, wear and tear ... of property used in the trade or business.” The applicable Treasury Regulations 111, § 29.23 (1) — 1, defines such allowance to be “that amount which should be set aside for the taxable year in accordance with a reasonably consistent plan . . . whereby the aggregate of the amounts so set aside, plus the salvage value, will, at the end of the useful life of the depreciable property, equal the cost or other basis of the property.” The Courts of Appeals have divided on the method of depreciation which is permissible in relation to such assets, and we therefore granted certiorari to resolve this conflict. 361 U. S. 810, 812. We have concluded that the reasonable allowance for depreciation of the property in question used in the taxpayer’s business is to be calculated over the estimated useful life of the asset while actually employed by the taxpayer, applying a depreciation base of the cost of the property to the taxpayer less its resale value at the estimated time of disposal.
In No. 143, Commissioner v. R. H. and J. M. Evans, the taxpayers are husband and wife. In 1950 and 1951, the husband, Robley Evans, was engaged in the business *94of leasing new automobiles to Evans U-Drive, Inc., at the rate of $45 per car per month. U-Drive in turn leased from 30% to 40% of the cars to its customers for long terms ranging from 18 to 36 months, while the remainder were rented to the public on a call basis for shorter periods. Robley Evans normally kept in stock a supply of new cars with which to service U-Drive and which he purchased at factory price from local automobile dealers. The latest model cars were required because of the demands of the rental business for a fleet of modern automobiles.
When the U-Drive service had an oversupply of cars that were used on short-term rental, it would return them to the taxpayer and he would sell them, disposing of the oldest and least desirable ones first. Normally the ones so disposed of had been used about 15 months and had been driven an average of 15,000 to 20,000 miles. They were ordinarily in first-class condition. It was likewise customary for the taxpayer to sell the long-term rental cars at the termination of their leases, ordinarily after about 50,000 miles of use. They also were usually in good condition. The taxpayer could have used the cars for a longer period, but customer demand for the latest model cars rendered the older styles of little value to the rental business. Because of this, taxpayer found it more profitable to sell the older cars to used car dealers, jobbers or brokers at current wholesale prices. Taxpayer sold 140 such cars in 1950 and 147 in 1951. On all cars leased to U-Drive, taxpayer claimed on his tax returns depreciation calculated on the basis of an estimated useful life of four years with no residual salvage value. The return for 1950, for example, indicated that each car’s cost to taxpayer was around $1,650;' after some 15 months’ use he sold it for $1,380; he charged depreciation of $515 based on a useful life of four years, without salvage value, which left him a net gain of $245, *95on which he calculated a capital gains tax. In 1951 the net gain based on the same method of calculation was approximately $350 per car, on which capital gains were computed. The Commissioner denied the depreciation claims, however, on the theory that useful life was not the total economic life of the automobile (i. e., the four years claimed), but only the period it was actually used by the taxpayer in his business; and that salvage value was not junk value but the resale value at the time of disposal. On this basis he estimated the useful life of each car at 17 months and salvage value at $1,325; depreciation was permitted only on the difference between this value and the original cost. The Tax Court accepted the Commissioner’s theory but made separate findings. The Court of Appeals reversed, holding that useful life was the total physical or economic life of the automobiles — not the period while useful in the taxpayer’s business. 264 F. 2d 502.
In No. 141, Massey Motors, Inc., v. United States, the taxpayer, a franchised Chrysler dealer, withdrew from shipments to it a certain number of new cars which were assigned to company officials and employees for use in company business. Other new cars from these shipments were rented to an unaffiliated finance company at a substantial profit.
The cars assigned to company personnel were uniformly sold at the end of 8,000 to 10,000 miles’ use or upon receipt of new models, whichever was earlier. The rental cars were sold after 40,000 miles or upon receipt of new models. For the most part, cars assigned to company personnel and the rental cars sold for more than they cost the taxpayer. During 1950 and 1951, the tax years involved here, the profit resulting from sale of company personnel cars was $11,272.80 and from rental cars, $525.84. The taxpayer calculated depreciation on the same theory as did taxpayer Evans, computing the gains on the sales at *96capital gain rates with a basis of cost less depreciation. The Commissioner disallowed the depreciation claimed. After paying the tax and being denied a refund, the taxpayer filed this suit. The trial court decided against the Commissioner. The Court of Appeals for the Fifth Circuit, however, reversed, sustaining the Commissioner’s views as to the meaning of useful life and salvage value. 264 F. 2d 552.
First, it may be well to orient ourselves. The Commissioner admits that the automobiles involved here are, for tax purposes, depreciable assets rather than ordinary stock in trade. Such assets, employed from day to day in business, generally decrease in utility and value' as they are used. It was the design of the Congress to permit the taxpayer to recover, tax free, the total cost to him of such capital assets; hence it recognized that this decrease in value — depreciation—was a legitimate tax deduction as business expense. It was the purpose of § 23 (1) and the regulations to make a meaningful allocation of this cost to the tax periods benefited by the use of the asset. In practical life, however, business concerns do not usually know how long an asset will be of profitable use to them or how long it may be utilized until no longer capable of functioning. But, for the most part, such assets are used for their entire economic life, and the depreciation base in such cases has long been recognized as the number of years the asset is expected to function profitably in use. The asset being of no further use at the end of such period, its salvage value, if anything, is only as scrap.
Some assets, however, are not acquired with intent to be employed in the business for their full economic life.. It is this type of asset, where the experience of the taxpayers clearly indicates a utilization of the asset for a substantially shorter period than its full eco*97nomic life, that we are concerned with in these cases. Admittedly, the automobiles are not retained by the taxpayers for their full economic life and, concededly, they do have substantial salvage, resale or second-hand value. Moreover, the application of the full-economic-life formula to taxpayers’ businesses here results in the receipt of substantial “profits” from the resale or “salvage” of the automobiles, which contradicts the usual application of the full-economic-life concept. There, the salvage value, if anything, is ordinarily nominal. Furthermore, the “profits” of the taxpayers here are capital gains and incur no more than a 25% tax rate. The depreciation, however, is deducted from ordinary income. By so translating the statute and the regulations, the taxpayers are able, through the deduction of this depreciation from ordinary income, to convert the inflated amounts from income taxable at ordinary rates to that taxable at the substantially lower capital gains rates. This, we believe, was not in the design of Congress.
It appears that the governing statute has at no time defined the terms “useful life” and “salvage value.” In the original Act, Congress did provide that a reasonable allowance would be permitted for “wear and tear of property arising out of its use or employment in the business.” (Emphasis added.) Act of Oct. 3,_ 1913, 38 Stat. 167. This language, particularly that emphasized above, may be fairly construed to mean that the wear and tear to the property must arise from its use in the business of the taxpayer — i. e., useful life is measured by the use in a taxpayer’s business, not by the full abstract economic life of the asset in any business. In 1918, the language of § 23 (1) was amended so that the words emphasized above would read “used in the trade or business,” § 214 (a)(8), Revenue Act of 1918, 40 Stat. 1067, and the section carried those words until 1942. Meanwhile, Treas. Reg. *9845, Art. 161, was promulgated in 1919 and continued in substantially the same form until 1941. It provided:
“The proper allowance for such depreciation of any property used in the trade or business is that amount which should be set aside for the taxable year in accordance with a consistent plan by which the aggregate of such amounts for the useful life of the property in the business will suffice, with the salvage value, at the end of such useful life to provide in place of the property its cost . . . .” (Emphasis added.)
It, too, may be construed to provide that the use and employment of the property in the business relates to the trade or business of the taxpayer — not, as is contended, to the type or class of assets subject to depreciation. The latter contention appears to give a strained meaning to the phrase. This might be particularly true of the language in Treasury Regulations 103, promulgated January 29, 1940, under the Internal Revenue Code of 1939. Its § 19.23 (1)-1 and § 19.23 (l)-(2) complement each *99other and seem to advise the taxpayer how to compute depreciation and what property is subject to it. The first section not only describes the proper allowance, but sets out how it is to be computed so that depreciation “plus the salvage value, will, at the end of the useful life of the property in the business, equal the cost . . . .” (Emphasis added.) The second section specifically defines the type of assets to which the depreciation allowance is applicable. It may be said that the taxpayers’ arguments as to this regulation fail completely, since it not only specifically provides that “useful life” relates to property while used “in the business,” but also details the type or class of property included within the allowance. It appears to-cut from under the taxpayers the argument that •the term “property used in the trade or business” relates to the type or class of assets that are included within the allowance. It would be strange to say that both of these sections of Regulations 103 defined the same thing, viz., the type or class of assets subject to depreciation. On the other hand, the taxpayers point out that Regulá-*100tions 111, issued in 1942, deleted the words “property in the business” from § 19.23 (1) — 1 and substituted the term “depreciable property.” This might, as taxpayers claim, establish that the phrase “property used in the trade or business” merely referred to the type of property involved. Certainly when considered in isolation, this appears to be true. But the “depreciable property” phrase does refer back to the earlier identical language, still remaining in the section, of “property used in the trade or business.” It does appear, however, as the Court of Appeals in No. 141, Massey, held, that this substitution was made because Congress expanded the depreciation allowance provision of § 23 (1) to include property held for the production of income. The change in the Regulations only conformed it to this amendment of the basic statute.
It is true, as taxpayers contend and as we have indicated, that the language of the statute and the regulations as we have heretofore traced them may not be precise and unambiguous as to the term “useful life.” It may be that the administrative practice with regard thereto may not be pointed to as an example of clarity, and that in some cases the Commissioner has acquiesced in inconsistent holdings. But from the promulgation of the first regulation in 1919, he has made it clear that salvage had some value and that it was to be considered as something other than zero in the depreciation equation. In fact many of the cases cited by the parties involved controversies over the actual value of salvage, not as scrap but on resale. The consistency of the Commissioner’s posi*101tion in this regard is evidenced by the fact that the definition of salvage as now incorporated in the regulations is identical with that claimed at least since 1941. In the light of this, it appears that the struggle over the term “useful life” takes on less practical significance, for, if salvage is the resale value and a deduction of this amount from cost is required, the dollar-wise importance to the taxpayer of the breadth in years of “useful life” is diminished. It is only when he can successfully claim that salvage means junk and has no value that an interpretation of “useful life” as the functional, economic, physical life of the automobile brings money to his pocket. Moreover, in the consideration of the appropriate interpretation of the term, it must be admitted that there is administrative practice and judicial decision in its favor, as we shall point out. Furthermore, as we- have said, Congress intended by the depreciation allowance not to make taxpayers a profit thereby, but merely to protect them from a loss. The concept is, as taxpayers say, but an accounting one and, we add, should not be exchangeable in the market place. Accuracy in accounting requires that correct tabulations, not artificial ones, be used. Certainly it is neither accurate nor correct to carry in the depreciation equation a value of nothing as salvage on the resale of the automobiles, when the taxpayers actually received substantial sums therefor. On balance, therefore, it appears clear that the weight of both fairness and argument is with the Commissioner.
Our conclusion as to this interpretation of the regulations is buttressed, we think, by a publication issued by the Commissioner in 1942, the same year as Regulations 111, and long before this controversy arose. It is known as Bulletin “F” and has been reissued as late as 1955. While it does not have the authority of a regulation, its significance is indicated clearly by the fact that both the taxpayers and the Commissioner point to it as conclusive *102of their respective views of the administrative practice. Likewise it is widely cited by tax authorities, as well as by the Courts of Appeals. A careful examination of the entire bulletin, however, indicates that it clearly supports the administrative practice claimed here by the Commissioner. For example, the title page warns that “[t]he estimated useful lives and rates of depreciation . . . are based on averages and are not prescribed for use in any particular case.” Again on page 2, Bulletin “F,” in discussing depreciation, emphasizes that it is based on “the useful life of the property in the business.” What is more significant is the simple clarity with which, on page 7, it defines salvage value to be “the amount realizable from the 'sale . . . when property has become no longer useful in the taxpayer’s business and is demolished, dismantled, or retired from service.” It even goes further to say that salvage “should serve to reduce depreciation, either through a reduction in the basis on which depreciation is computed or a reduction in the rate.”
Moreover, Congress was aware of this prior prevailing administrative practice as well as the concept of depreciation upon which it was based. Although the tax years involved here are 1950 and 1951, we believe that the action of Congress in adopting the 1954 Code should be noted, since it specifically recognized the existing depreciation equation. For the first time, the term “useful life” was inserted in the statutory provision. The accompanying House Report to the bill stated:
“Depreciation allowances are the method by which the capital invested in an asset is recovered tax-free over the years it is used in a business. The annual *103deduction is computed by spreading the cost of the property over its estimated useful life.” H. R. Rep. No. 1337, 83d Cong., 2d Sess. 22.
It is also noteworthy that the report states that “The changes made by your committee’s bill merely affect the timing and not the ultimate amount of depreciation deductions with respect to a property.” Id., at 25.
Moreover, as we have said, there are numerous cases in the Tax Court in which depreciation was permitted only on the useful life of the property in the taxpayer’s business. The taxpayers point to others which appear to be to the contrary. In most of these, however, the issue was factual, i. e., the time lapse before the property would wear out from use or, as we have said, its salvage or resale value. They cannot be said to prove conclusively that the Commissioner was following a physically useful-life theory; for there is no affirmative showing or finding as to the length of the physically useful life. The most that can be said is that the element of compromise probably played a predominant role in the result in each case. Moreover, there is no indication in any of these cases that the amount of depreciation would have been changed *104by computing it on the basis of its useful life in the business. The cases do not seem to reflect considered judgments as to the proper meaning of the terms used in the depreciation equation and we find them of little value as precedents.
Finally, it is the primary purpose of depreciation accounting to further the integrity of ^periodic income statements by making a meaningful allocation of the cost entailed in the use (excluding maintenance expense) of the asset to the periods to which it contributes. This accounting system has had the approval of this Court since United States v. Ludey, 274 U. S. 295, 301 (1927), when Mr. Justice Brandéis said, “The theory underlying this allowance for depreciation is that by using up the plant, a gradual sale is made of it.” The analogy applies equally to automobiles. Likewise in Detroit Edison Co. v. Commissioner, 319 U. S. 98, 101 (1943), this Court said:
“The end and purpose of it all [depreciation accounting] is to approximate and reflect the financial consequences to the taxpayer of the subtle effects of time and use on the value of his capital assets. For this purpose it is sound, accounting practice annually to accrue ... an amount which at the time it is retired will with its salvage value replace the original investment therein.”
Obviously a meaningful annual accrual requires an accurate estimation of how much the depreciation will total. The failure to take into account a known estimate of salvage value prevents this, since it will result in an understatement of income during the years the asset is employed and an overstatement in the year of its disposition. The practice has therefore grown up of subtracting salvage value from the purchase price to determine the *105depreciation base. On the other hand, to calculate arbitrarily the expected total expense entailed by the asset on the false assumption that the asset will be held until it has no value is to invite an erroneous depreciation base and depreciation rate, which may result in either an over- or an under-depreciation during the period of use. If the depreciation rate and base turn out to reflect the actual cost of employing the asset, it will be by accident only. The likelihood of presenting an inaccurate picture of •yearly income from operation's is particularly offensive where, as here, the taxpayers stoutly maintain that they are only in the business of renting and leasing automobiles, not of selling them. The alternative is to estimate the period the asset will be held in the business and the price that will be received for it on retirement. Of course, there is a risk of error in such projections, but prediction is the very essence of depreciation accounting. Besides, the possibility of error is significantly less where probabilities rather than accidents are relied upon to produce what is hoped to be an accurate estimation of the expense involved in utilizing the asset. Moreover, under a system where the real salvage price and actual duration of use are relevant, to further insure a correct depreciation base in the years after a mistake has been discovered, adjustments may be made when it appears that a miscalculation has been made.
*106Accounting for financial management and accounting for federal income tax purposes both focus on the need for an accurate determination of the net income from operations of a givén business for a fiscal period. The approach taken by the Commissioner computes depreciation expense in a manner which is far more likely to reflect correctly the actual cost over the years in which the asset is employed in the business.
*107We therefore conclude that the Congress intended that the taxpayer should, under the allowance for depreciation, recover only the cost of the asset less the estimated salvage, resale or second-hand value. This requires that the useful life of the asset be related to the period for which it may reasonably be expected to be employed in the taxpayer’s business. Likewise salvage value must include estimated resale or second-hand value. It follows that No. 141, Massey Motors, Inc., v. United States, must be affirmed, and No. 143, Commissioner v. R. H. and J. M. Evans, reversed.
It is so ordered.
Mr. Justice Harlan,
whom Mr. Justice Whittaker, and Mr. Justice Stewart join, dissenting in Nos. 141 and 143, and concurring in the judgment in No. 283.
This is one of those situations where what may be thought to be an appealing practical position on the part of the Government has obscured the weaknesses of its legal position, at least in Nos. 141 and 143.
The position which the Commissioner takes in these cases with respect to the basic issue of “useful life” is that contained in the regulations promulgated by him in 1956 under the Internal Revenue Code of 1954, which define the useful life of a depreciable asset as the
“period over which the asset may reasonably be expected to be useful to the taxpayer in his trade or business . ...”
In No. 283 the Commissioner seeks to apply this regulatory definition to the returns of the taxpayer with respect to the taxable years ended March 31,1954,1955, and 1956. *108In Nos. 141 and 143 he seeks in effect to apply the same definition to the taxable years 1950 and 1951, both of which were of course long before the enactment of the 1954 Code. See 264 F. 2d, at 506.
I agree that these regulations represent a reasonable method for calculating depreciation within the meaning of the 1954 Code, and that they are valid as applied prospectively. But since I believe that as to “useful life” they are wholly inconsistent with the position uniformly taken by the Commissioner in the past, I do not think they can be applied retrospectively in all instances. While I consider that the regulations may be so applied in No. 283, in my opinion that is not so in Nos. 141 and 143.
I.
It is first important to understand the precise nature of the issues before the Court. Both the method of depreciation contended for by the taxpayers and that urged by the Government purport to allocate an appropriate portion of an asset’s total cost to each o.f the years during which the taxpayerholds it. Both methods define the total cost to be so allocated as the original cost of the asset less its salvage value at the end of its useful life. And under both methods, the total cost to be allocated is divided by the number of years in the useful life and the resulting figure is deducted from the taxpayer’s income each year he holds the asset. As the Court correctly notes, the practical difference in the end results of the two methods involves the extent to which a taxpayer may be able to obtain capital-gains treatment for assets sold at or before the end of their useful life for amounts realized in excess of their remaining undepreciated cost.
The difference between the two methods from a theoretical standpoint is simply this: The taxpayers define useful life as the estimated physical life of the *109asset, while the Government defines the term as the period during which the taxpayer anticipates actually retaining the asset in his business. Thus, under the taxpayers’ system, the total cost to be allocated is original cost less the salvage or junk value of the asset at the end of its physical life. This figure is divided by the number of years of estimated physical life, and the quotient is subtracted from income each year the taxpayer holds the asset. Under the Government’s method, the total cost to be allocated is original cost less the “salvage” value at the end of the asset’s actual use in the business, that is, less the price anticipated on its resale at that time, even though the asset may not be in fact physically exhausted. This figure is divided by the number of years in the holding period, and the quotient is subtracted from income each year the taxpayer holds the asset.
If an asset is held until it is physically exhausted, both methods produce exactly the same result. Similarly, both methods can result in inaccuracies if predictions of useful life and salvage value turn out to be wrong. The Government, however, contends that where it can be predicted with reasonable certainty that an asset will be disposed of before the end of its physical life, its method of depreciation is more likely to reflect the true cost of the asset to the particular business. This is said to be so because the true cost to the business, in the end, is the asset’s original cost less the amount recovered on its resale, and the Government’s method starts from an estimate of that amount, which is then allocated among the years involved. The taxpayers’ method on the other hand, starts from an estimate of the end cost of the asset in the general business world, and will accurately reflect such cost to the taxpayer’s business only if the decline in market value at the time of resale can be expected to correspond roughly to the portion of the asset’s general business end *110cost which has been theretofore depreciated. In many cases that may be true, but in the present cases, there is in fact a great disparity between actual decline in market value at the time of resale and the portion of cost theretofore depreciated under taxpayers’ method.
It need not be decided whether, as an abstract matter, one method or the other is deemed preferable in accounting practice. Apparently there is a split of authority on that very question. It is sufficient to note that in most instances, either method seems to give satisfactory results. Assuming that because of the unusual case, such as we have here, the Government’s method on the whole may more accurately reflect the cost to a particular taxpayer’s business, the question for me is whether the Commissioner has nevertheless established a practice to the contrary upon which taxpayers were entitled to rely until changed by him. I turn now to the examination of that question.
h-i HH
The Court relies on the wording of certain revenue statutes and regulations to show that the period during which depreciable assets are employed in the taxpayer’s business, as opposed to the period of their physical life, has always been regarded as useful life for purposes of depreciation. Concededly, the term useful life did not appear in the statute until the Internal Revenue Code of 1954, and though it had appeared in the regulations as early as 1919, Treas. Reg. 45, Art. 161, was never defined therein until 1956, ante, p. 107, when the Commissioner took the position he now asserts. The Court seizes on language which was not directed to the present problem and which could equally be read to support the *111Government’s or the taxpayers’ contention. The situation before 1956 was as follows:
The Act of Oct. 3, 1913, permitted a reasonable allowance for “wear and tear of property arising out of its use or employment in the business.” It is certainly true, as the Court says, that this means that “the wear and tear to the property must arise from its use in the business of the taxpayer.” But it does not follow at all that the formula for calculating that wear and tear must be based on a useful life equal to the period the asset is held in the business. For, as noted above, a formula based on the physical life of the asset also results in an estimate of the portion of the asset’s total cost attributable to its use in the business, and may in some circumstances yield the same tax consequences as a “holding-period” formula.
Treasury Regulations 45, Art. 161, promulgated in 1919 and continued in substantially the same form until 1942, provided that the taxpayer should set aside each year an amount such that “the aggregate of such amounts for the useful life of the property in the business will suffice, with the salvage value, at the end of such useful life to provide in place of the property its cost . . . .” In 1942, the statute was amended to permit depreciation, not only, as before, on property used in the trade or business, but also on property held for the production of income. Accordingly, the regulation was revised to delete the words “property in the business” and substitute therefor “the depreciable property.” Reg. Ill, § 29.23 (Z) — 1. The Court says that the deleted term could not have been meant to define the type of property subject to the depreciation allowance, since that function was already performed by another section of the regulation. That may be true, but it does not show that the language was meant to define the period of useful life.- If it had been so meant, the Commissioner *112would hardly have simply substituted “useful life of the depreciable property” for “useful life of the property in the business,” but would have inserted appropriate language, such as “useful life of the property while used in the business or held for the production of income.” It is quite evident that the question of a holding period different from the physical life of the property was never adverted to, and that the term “property in the business,” while not an affirmative definition of the type of property subject to depreciation, simply referred to that definition in connection with useful life because it was apparently assumed that assets were generally held in a taxpayer’s business until worn out.
In light of the above, the Government’s reliance on cases such as United States v. Ludey, 274 U. S. 295, 300-301, and Detroit Edison Co. v. Commissioner, 319 U. S. 98, 101, is wide of the mark. The language relied upon in Ludey is virtually identical to that contained in the pre-1942 regulations, and that in Detroit Edison merely says that the purpose of depreciation is to recover, by the time of an asset’s retirement, the original investment therein. As noted above, depreciation based on either definition of useful life is dedicated to that end. The Government’s reliance on Bulletin “F” is also misplaced. The Court refers to a statement on page 2 of the Bulletin which merely lifts from the regulation the phrase “useful life of the property in the business.” The Court also relies on a statement appearing on page 7, defining salvage as. “the amount realizable from the sale . . . when property has become no longer useful in the taxpayer’s business and is demolished, dismantled, or retired from service.” (Emphasis added.) The italicized language again reveals the assumption that assets were generally intended for use in the business until their physical exhaustion. The present question was never adverted to.
*113I believe, therefore, that the statute and regulations are wholly inconclusive, and' that the Commissioner’s position can be gleaned only from the stand he has taken in litigated cases. I turn now to those cases. Contrary to the picture of uncertainty which the Court draws from them, I believe they leave little room for doubt but that the Commissioner’s pre-1956 position on “useful life” was flatly opposed to that which he now takes.
III.
In examining the cases, it must be borne in mind that even the Commissioner does not contend that a taxpayer who happens to dispose of some asset before its physical exhaustion must depreciate it on a, useful life equal to the time it was actually held. It is only when the asset “may reasonably be expected” to be disposed of prior to the end of its physical life that the taxpayer must base depreciation on the shorter period. Reg. § 1.167 (a)-l (b). Therefore, the only cases relevant in this regard are those in which the taxpayer’s past experience indicated that assets would be disposed of prior to becoming junk, thus presenting the issue whether the shorter or longer period should control for purposes of depreciation.
In four such cases, involving tax years prior to 1942, the taxpayer had a practice of disposing of assets substantially prior to their physical exhaustion. In Merkle Broom Co., 3 B. T. A. 1084, the taxpayer customarily disposed of its automobiles after two years. It attempted to depreciate them over a three-year useful life; the Com*114missioner asserted a five-year useful life; and the court allowed four years.
In Kurtz, 8 B. T. A. 679, the taxpayers customarily sold their automobiles after two or three years at substantial values. They depreciated on a four-year useful life; the Commissioner asserted a five-year life; and the court agreed.
In Sanford Cotton Mills, 14 B. T. A. 1210, the taxpayer customarily disposed of its motor trucks after two and one-half years. It claimed a three-year useful life; the Commissioner asserted a five-year useful life; and the court found that four years was reasonable.
In General Securities Co., 1942 P-H BTA-TC Mem. Dec. ¶ 42,219, the taxpayer sold its automobiles after one or two years. The court held that a reasonable useful life was three years.
It is apparent from these cases that both the Commissioner and the courts were thinking solely in terms of the physical life of the asset, despite the fact that the taxpayer customarily held the assets for a substantially shorter period. In at least some of the cases, it would have made a very real difference had depreciation been calculated on the basis that the useful life of the asset meant its holding period. For example, in the Sanford case, taxpayer’s trucks were sold after two and one-half years at less than one-seventh of their original cost. Given the five-year useful life proposed by the Commissioner, taxpayer would have had, at the time of resale, an undepreciated basis equal to half the original cost, while the proceeds of resale would have brought it only one-seventh of original cost, thus giving rise to a loss of the difference. If the Government’s present position had been applied, the difference between original cost and resale value would have been depreciated over two and one-half years, giving rise to no gain or loss at the end of that time. Similarly, in the General Securities case, given a three-year useful life, the *115taxpayer’s automobiles, when traded in after one year, had an undepreciated basis of two-thirds of original cost, yet their resale brought only one-half to one-third of their original cost, again resulting in a substantial loss which would have been avoided under the Government’s present method.
It is true that the only tax distortion present in these cases was a shift of ordinary deductions from the years in which the property was used in the business to the final year of its disposition. It is also true that had the situation been reversed, so that depreciation on a physical-life basis outran decline in market value, the resulting gain in the year of disposition would have been ordinary income, since capital-gains treatment for disposition of property used in the trade or business was not accorded by Congress until 1942. However, it is significant that the Commissioner’s adherence to a physical-life method did result in a distortion of income by shifting deductions among various tax years, which often entails serious revenue consequences, and that by 1942 physical life seems to have been uniformly accepted as the proper definition of useful life.
In light of these circumstances, four cases involving tax years subsequent to 1942 acquire special significance. In Pilot Freight Carriers, Inc., 15 CCH T. C. Mem. 1027, the taxpayer disposed of its tractors after an average of 38 months and its trailers after an average of 32.6 months. It claimed depreciation on a four-year useful life with 10% or less salvage value. The Commissioner asserted useful lives of five and six years for the tractors and trailers, respectively, and the court found that four and five years, respectively, was reasonable. It is to be noted that upon resale, taxpayer received, because of wartime inflation, amounts substantially in excess of undepreciated *116cost, resulting in large capital gains. Yet the Commissioner, in attempting to correct this disparity, asserted only that useful life should be increased to reflect more accurately the physical exhaustion of the assets, not that it should be equated with the holding period.
In Lynch-Davidson Motors, Inc., v. Tomlinson, 58-2 U. S. T. C. ¶ 9738, an automobile dealer disposed of company cars each year when new models were brought out, yet depreciated on a three-year useful life with salvage value of $50. The Commissioner did not dispute this method of depreciation and the court held it to be proper. In the companion case of Davidson v. Tomlinson, 58-2 U. S. T. C. ¶ 9739, taxpayers were in the automobile rental business, and kept their automobiles only one year. They also were permitted to depreciate on a useful life of three years with $50 salvage value. The striking similarity between the facts of these two cases and those of the present ones need not be elaborated.
Finally, as late as 1959, in Hillard, 31 T. C. 961, the Commissioner took the position that the taxpayer, who operated a car rental business, and who disposed of his cars after one year, should depreciate them on the basis of a four-year useful life rather than the three years contended for by taxpayer.
Thus in all these cases, as in the cases before us, the problem of offsetting depreciation deductions by capital gains existed; nevertheless the Commissioner consistently adhered to the position, adopted long prior to 1942, that physical life controlled.
The Court, however, seems to believe that the effect of these cases is vitiated by several cases dealing with “salvage” value. In three of such cases, the assets were *117apparently held by the taxpayer until at or near the end oí their physical lives, and the only issue was whether the taxpayer had erroneously calculated the salvage value at the end of that time. Thus they are of no significance for present purposes.
The Court’s view fares no better under any other of these cases. In Bolta Co., 4 CCH T. C. Mem. 1067, involving a 1941 tax year, the taxpayer disposed of several machines after they had ceased to be useful in its business but while they were still useful in other businesses. It projected an average holding period of five years and assumed no salvage value. The Commissioner acquiesced in the five-year useful life but contended that the taxpayer could reasonably have anticipated a salvage value equal to 25% of original cost. The court agreed.
In Koelling v. United States, 57-1 U. S. T. C. ¶ 9453, taxpayers disposed of cattle after they were no longer useful for breeding, and depreciated them on a useful life equal to that period, making no allowance for salvage value. The Commissioner found that it was unreasonable thus to deduct the entire cost of the animals over their breeding life, and required the taxpayers to deduct as salvage value their predicted resale price.
In Cohn v. United States, 259 F. 2d 371, taxpayers had established flying schools during 1941 and 1942 under contract with the Army Air Corps. The arrangement was expected to last only until the end of 1944, and the useful fife of property used in the business was calculated on that basis, with no allowance for salvage value. The Commissioner asserted various longer useful lives for the property, varying from five to ten years. The court permitted the taxpayers to use the shorter useful life, but required them to deduct the reasonable salvage value of the equipment which would be realized at the end of that period. The Government did not appeal from the useful-*118life ruling and the only dispute was over the correct amount of salvage value.
Thus in two of the relevant salvage-value cases, Bolta and Koelling, the taxpayer himself proposed a useful life equivalent' to holding period but employed a hybrid version by failing to adopt the corresponding concept of salvage value. The Commissioner merely took the position that if the holding-period method was to be used, it must be used consistently by deducting the appropriate salvage value. In the third, Cohn, the Commissioner actually rejected the taxpayers’ attempt to employ the holding period and merely acquiesced when the court permitted the taxpayers to do so, provided the corresponding salvage value was deducted. However, in no case, until the present ones, does it appear that the Commissioner has ever sought to require the taxpayer to use the holding-period method where the taxpayer has attempted to use physical life. And I do not understand the Government to controvert this. To the contrary, the Commissioner has not infrequently required the taxpayer to depreciate on the basis of physical life where the taxpayer had attempted to employ a shorter period, even in instances where significant capital-gains consequences turned on the difference. Indeed, as the Lynch-Davidson, Davidson, and Hillard cases, supra, indicate, the Commissioner, until quite recently, has adhered to the physical-life concept in automobile cases virtually indistinguishable from the present ones. In the past the Commissioner, unsuccessfully, has merely sought to curb the capital-gains possibilities in such instances by contending that the automobiles involved were not depreciable assets subject to capital-gains treatment under § 117 (j) of the Internal Revenue Code of 1939. Having conceded that the property involved in the present cases is subject to the depreciation deduction, I do not think the Commissioner should now be permitted to defeat his own position as regards the *119meaning of “useful life” — a position consistently maintained by him over a period of 33 years from 1926 to 1959 in every litigated case to which our attention has been called — by requiring these taxpayers, in respect of taxable years not subject to the provisions of the 1954 Code, to adopt a holding-period formula for useful life in depreciating the assets in question. Cf. Helvering v. R. J. Reynolds Tobacco Co., 306 U. S. 110, and Helvering v. Griffiths, 318 U. S. 371. In the application of this salutary principle it should make no difference that the Commissioner’s earlier different practice was not embodied in a formal regulation. Cf. Helvering v. Reynolds, 313 U. S. 428, 432; Higgins v. Commissioner, 312 U. S. 212, 216.
Accordingly, I would reverse in No. 141 and affirm in No. 143.
IV.
The situation presented in No. 283 is, however-, different. The taxable years in question there are those terminating on March 31, 1954, 1955, and 1956, respectively. All the taxable years thus ended before the promulgation of the new depreciation regulations on June 11, 1956. The Government concedes that Congress did not change the concept of useful life when it enacted the 1954 Code. Therefore, the question here is whether the Commissioner can, by a formal regulation, change his position retroactive only to the effective date of the statute under which it is promulgated.
Petitioner, relying on Helvering v. R. J. Reynolds Tobacco Co. and Helvering v. Griffiths, supra, asserts that where a regulation interpreting a statute has been in force for some time and has survived the re-enactment of the statute, the Commissioner cannot retroactively change *120that interpretation by a new regulation. However, here the Commissioner’s earlier adherence to the physical-life concept of useful life was expressed not in the regulations — which did not refer to the problem — but in his own administrative practice. Therefore, the present case is more like Helvering v. Reynolds, 313 U. S. 428, wherein this Court permitted the Commissioner to apply a regulation retroactive to the effective date of the statute under which it was promulgated, where his previous contrary position had been expressed only by informal administrative practice, even though the statute had been re-enacted in the interim. Application of this principle in the present case is the more called for, since Congress, in the 1954 Code, has for the first time used the term “useful life” and has made the availability of certain new accelerated methods of depreciation — among them the so-called “declining balance method,” used by the taxpayer here — dependent upon its definition. It is appropriate therefore to permit the Treasury maximum discretion in integrating the concept of useful life into the new provisions and in doing so from the effective date of the statute forward.
Since the statute permits use of the declining-balance method only as to property with a useful life of three years or more, it follows that the Commissioner properly disallowed use of the declining-balance method as to Hertz’ automobiles, whose useful life under the new regulation was less than three years. As to its trucks, admittedly held for more than three years, the only remaining question is whether Hertz should be allowed to depreciate them below what the Commissioner considers to be a reasonable salvage value. Given the fact that the Commissioner’s definition of salvage value as resale price on disposition of the asset at the end of its holding period is validly applicable to Hertz, it becomes important that the declining-balance method not be construed to defeat *121that concept. Were there no “salvage stop” in connection with declining-balance depreciation, it is clear that taxpayers who held assets for relatively short periods of time might be able to depreciate far below anticipated resale price, since the declining-balance rate is applied against the entire cost of the asset undiminished by salvage. Since the legislative history of the statute in this regard is ambiguous at best, and since there is no prior statute or administrative interpretation to bedloud the issue, the Commissioner’s construction should be allowed to stand. Accordingly, I concur in the Court’s judgment affirming No. 283.
Mr. Justice Douglas
joins Parts I, II, and III of this opinion. He would, however, reverse in No. 283 — Hertz Corp. v. United States, on the ground that the change in administrative practice involved here should not be retroactively applied under the circumstances of this case. Cf. United States v. Leslie Salt Co., 350 U. S. 383, 396.