This is a consolidated appeal by the Department of Energy (DOE) from summary judgments in favor of 15 oil refiners, granted by the District Court for the Northern District of Ohio, Eastern Division, and the District Court for the District of Delaware. Nine suits were filed and consolidated for trial in Ohio and six in Delaware. All involved the interpretation of petroleum price control regulations issued by the Federal Energy Administration (FEA) governing the recovery by oil refiners of two categories of costs for the period January 1, 1975 through January 31, 1976. The FEA interpretation of cost pass through regulations required refiners to allocate monthly sales revenues first to the recoupment of all increased “product costs” (primarily the costs of crude oil and purchased petroleum products) and then to the recoupment of “non-product” costs (primarily operating and marketing costs).1
I. PROCEEDINGS IN DISTRICT COURT
The suits, challenging the FEA’s interpretation of the pass through regulations and seeking declaratory and injunctive relief, were filed against the FEA2 after it declined to grant the plaintiff refiners a class exception from the impact of its ruling and required the refiners instead to seek individual exception relief. Both courts denied the FEA’s motions to dismiss on ripeness and exhaustion grounds and held in essence that the following legal questions were appropriate for immediate consideration: (1) whether the FEA’s interpretation of the meaning of the regulations governing refiner pass through of increased costs was (a) correct and (b) required; and (2) if so, whether the regulations were invalid because of failure to comply with the procedural and substantive statutes pursuant to which the regulations were promulgated.3
*1034All parties in both cases filed cross-motions for summary judgment. The primary issue before each court was the validity of FEA’s ruling that refiners were required to recover product cost increases first.4 During the relevant period (i. e., January 1, 1975 through January 31, 1976) refiners used three methods of allocating cost recoveries: (1) non-product cost increases recovered first (NPCI First); (2) product cost increases recovered first (NPCI Last); and (3) the “proportional method”, product and non-product increases recovered on a prorated basis. The principal distinction between the Ohio and Delaware cases is that all of the Delaware plaintiffs recovered their product and non-product cost increases under the proportional method, while a majority of the Ohio plaintiffs used the non-product cost increases first method.5
In granting summary judgment in Standard Oil Company v. Federal Energy Administration (Standard Oil II), 453 F.Supp. 203, 245-46 (1978), the Ohio court held that (1) the “FEA never issued a valid all product costs first passthrough rule applicable to the time frame January 1, 1975 to January 31,1976”; (2) the “FEA never issued an all product costs first sequence rule . . . in compliance with the required rulemaking procedures and therefore any such rule is void”; (3) the “FEA did not issue the explicit rule banning the banking of non-product costs in compliance with the required rulemaking procedures”; but (4) the all product costs first rule was within the FEA’s statutory authority. The court also found constitutional questions raised by ap-pellees were substantial and certified those issues to this court.
In Phillips Petroleum Co. v. Department of Energy (Phillips Pet. II), 449 F.Supp. 760, 801-02 (1978), the Delaware court held that (1) the regulations affecting prices in effect before December 1, 1974 “did not contain a tacit prohibition against NPCI banking” of non-product cost increases and “did not implicitly require the use of NPCI Last method of cost recovery”; (2) “even if the regulations reasonably could have been interpreted to require that NPCI be recovered last, they also could have been interpreted to require that product and non-product cost increases be recovered proportionally”; (3) “the FEA did not decide to require use of the NPCI Last method until the very end of the relevant period”, did not announce its decision until February 1, 1976, and this belated interpretation “did not qualify for retroactive application”; and (4) the FEA did not comply with the notice and comment provisions of the Administrative Procedure Act and the Federal Energy Administration Act in its purported adoption of either the prohibition of NPCI banking or the NPCI Last requirement.
Preliminary to a consideration of the contentions of the respective parties on this appeal, we shall set forth in summary form *1035the statutory and regulatory background6 and, in particular, the procedures in FEA’s promulgation of the sequence of recovery rule upon which it relies in these cases.
II. THE STATUTORY AND REGULATORY BACKGROUND
The petroleum pricing regulations in effect for the relevant period were derived from rules promulgated by the Cost of Living Council (CLC) under the authority of the Economic Stabilization Act of 1970, as amended, 12 U.S.C. § 1904 note §§ 201-220 and Phase IV of the Economic Stabilization Program. In July of 1973, the CLC published proposed general rules for “Phase IV” price controls to become effective at the end of a 60 day freeze imposed by executive order on June 13, 1973. Special rules — to be included as Subpart L of the regulations — were proposed for the petroleum industry. Under the proposed special rules, maximum lawful prices for refined petroleum products would consist of (1) the May 15, 1973 price to a class of customers, (2) increased costs of domestic crude oil and imports, subject to a profit margin limitation, and (3) other allowable costs, subject to pre-notification and a profit margin limitation.7
On August 19, 1973 the CLC added Sub-part L — Petroleum and Petroleum Products — to the Phase IV regulations, which established special rules governing the pricing of petroleum products. The key to this pricing structure was the “base price”, which was defined as the May 15, 1973 selling price plus adjustments for the increased costs of imports and domestic crude petroleum above the May 15 price.8 This definition, which differed substantially from that in the proposed regulations, allowed refiners to pass through their product cost increases on a dollar for dollar basis without being subject to pre-notification requirements or a profit margin rule. As to non-product costs, refiners could make price increases above base price to reflect these costs, but they remained subject to pre-noti-fication and profit margin limitations.
As initially promulgated, the petroleum pricing regulations did not provide that either product cost increases or non-product cost increases unrecouped in one month could be carried forward, or “banked”, for use in the pricing equation for establishing prices in a later month. In September, 1973, the CLC amended its regulations to provide that product cost increases incurred in one month which were not recouped in the following month, could be recovered in a later month. § 150.356(c)(1); 38 Fed.Reg. 25686, 25688 (Sept. 14, 1973).9 There was no express provision allowing or prohibiting *1036the banking of non-product increases until November, 1974, when banking non-product cost increases was specifically prohibited. The only public explanation for the special treatment accorded product cost increases was a statement in a press release issued by Dr. John Dunlap, Chairman of the CLC, on August 10, 1973:
The Council has been very concerned that the final regulations strike a delicate balance between constraining prices while at the same time encouraging the necessary increase in supplies which the country must have. The Council is aware that energy prices must be allowed to rise in order to stimulate development of new energy reserves and make possible the purchase of higher cost foreign oil. At the same time, we must prevent unnecessary price increases.
Phillips Pet. II, 449 F.Supp. at 771-72.
In November, 1973, Congress passed the Emergency Petroleum Allocation Act (EPAA), Pub.L. 93-159, 87 Stat. 627 (Nov. 27, 1973), 15 U.S.C. §§ 751 et seq. On December 4,1973, the President established the Federal Energy Office (FEO) and delegated to it the authority to implement the allocation and price stabilization provisions of the EPAA.10 The FEO adopted the CLC’s Phase IV petroleum pricing regulations without modification.11 On May 7, 1974 the Federal Energy Administration Act (FEAA) was signed into law. Pub.L. 93-275, 88 Stat. 96; 15 U.S.C. §§ 761 et seq. In June, 1974, the President abolished the FEO, replaced it with the Federal Energy Administration (FEA) and delegated to it the authority previously delegated to the FEO.12
On September 4, 1974, the FEA gave notice of its intent to revise the petroleum price regulations to accomplish three objectives: (1) to simplify and clarify the price regulations, while retaining the essential concepts of the regulations originally promulgated by the CLC; (2) to eliminate unnecessary regulatory restrictions; and (3) to help restore competition so that traditional market forces could begin to supplant the rigid price control regulations as the price setting mechanism. To accomplish these objectives the FEA proposed to amend the regulations to eliminate the pre-notification procedure and replace it with an automatic pass through for certain non-product cost increases similar to the procedure used for pass through of increased product costs; to provide more flexibility in the manner in which increased product costs could be applied by refiners; and to limit the extent to which increased product costs unrecouped in one month could be recouped in subsequent months. 39 Fed.Reg. 32718 (September 10, 1974).
After taking oral and written comments, the FEA on November 1, 1974, issued the first of its amendments pursuant to the September 6 notice. Instead of adopting a virtual phase out of the increased product costs banking provisions as it initially contemplated, the FEA modified its position in response to opposition by the refiners and resellers and retailers. It recognized that phase out of the banking system would have had a “modest inflationary effect” by inducing sellers to recover the maximum amount of their increased product costs as soon as possible. In contrast, the banking provisions had given refiners a mechanism with which to smooth out price adjustments over a period of time. Still the FEA was concerned with the large amounts of increased product costs which refiners had banked. If a petroleum shortage occurred, pass through of the refiners’ entire bank of increased product costs at once, which was allowable under current regulations, could result in drastic price increases. Consequently, the FEA amended the regulations to restrict the amount of banked product costs which could be used to increase the *1037base price in a particular month to 10 percent of the total amount of unrecouped increased product costs available for pass-through. § 212.83(d)(3); 39 Fed.Reg. 39259, 39260-61 (November 6, 1974).
On November 29, 1974, the FEA issued further amendments to the petroleum pricing regulations. Among other things it deleted the pre-notification procedure for non-product cost increases, adopted a more specific definition of non-product costs, and made clear that the increased cost of crude oil consumed as a refinery fuel, that is oil used in the refinery processes, was a non-product cost. 39 Fed.Reg. 42368 (December 5, 1974). Additionally, it for the first time adopted an explicit rule that increased non-product costs not recouped in the current month could not be carried forward for use in computing allowable prices in excess of base prices in any subsequent month. § 212.83(e)(4); 39 Fed.Reg. at 42372. These amendments took effect on January 1,1975.
III. PROMULGATION OF SEQUENCE OF RECOVERY RULE
During the initial phases of petroleum price and allocation regulation by the CLC there was apparently little, if any, concern about the problem of sequence of cost recovery. David G. Wilson, Deputy General Counsel of the FEA, stated in his deposition that the draftsmen of the price regulations had not thought about the sequence question, nor had they “focuse[d] on exactly how costs would be treated as recouped in prices”. Wilson Depo. 40-41. (R. 1113-14). The draftsmen were concerned with the process of computing the base price and the maximum selling price, not with the process of accounting for and allocating sales revenues in the event less than the maximum price was being charged. There was no explicit sequence of recovery rule.
Following the adoption of the amendments effective January 1, 1975, the refiners, as noted supra, used three methods of recovering increased costs: (1) the “NPCI First” method, where non-product cost increases were recovered first; (2) the “NPCI Last” method, where product cost increases were first recovered; and (3) the “Proportional” method, where cost increases were recovered pro rata.13 Between January 1, 1975, and January 31, 1976, several FEA officials, through instruction manuals, directives, forms and worksheets issued to FEA auditors and disseminated to refiners, indicated that the FEA construed its regulations to permit or require the use of the proportional method.14
On February 1, 1976, the FEA promulgated amendments to its petroleum price regulations, effective on that date. One of the new regulations (10 C.F.R. § 212.85) explicitly required refiners to use the NPCI Last method to calculate recoupment of increased costs. 41 Fed.Reg. 5111, 5113, 5120 (February 4, 1976). In the preamble the FEA asserted that this section entailed no change in the regulatory framework. Although it pointed out that the regulations “were silent as to the order in which various categories of increased costs would be *1038deemed to have been recouped by refiners in their prices for covered products,” the FEA claimed that in essence “the order specified in the new § 212.85 is the same as that under the regulations previously in effect”. Id. at 5113.
In numerous comments following the promulgation of this rule refiners objected to its application both prospectively and retroactively. The comments were nearly unanimous that the rule was a drastic change from former practice and would result in rather dire consequences. In April, 1976, after providing notice and holding public hearings, the FEA revoked retroactively to February 1, 1976 the regulation requiring use of the NPCI Last method and also the regulation prohibiting the banking of non-product cost increases. The FEA concluded that the combination of the ban on banking NPCI and the NPCI Last sequence of recovery rule would cause the following undesirable effects: (1) inflation; (2) “prices would . . . tend to wide monthly fluctuations”, (3) “a disincentive for refiners to build up inventories”, (4) an incentive to “decrease refinery production”, and (5) reconsideration, deferment, or even elimination of “capital investment to expand refinery capacity”. 41 Fed.Reg. 15330, 15331 (1976). The rules would compel refiners to recover all costs currently to minimize absorption and would lead them to rely more on imported, purchased products, since refinery production cost increases (labor, fuel, etc.) could not be banked.
Despite its findings with respect to the deleterious effects which would result from a prospective application of the NPCI Last rule combined with the ban on NPCI banking, the FEA continued to maintain that the NPCI Last sequence had been implicit in the price regulations effective for the period from January 1, 1975 through January 31, 1976. On August 3, 1976, however, the FEA proposed to grant a “class exception” to permit all refiners who were not constrained by the profit margin limitation to recompute their increased cost recoveries during the relevant period using “the ‘proportional’ cost recovery approach”, 41 Fed. Reg. 33282, 33283 (August 9, 1976). The FEA noted that “many — perhaps even a majority of — refiners” had used either the proportional or NPCI First method and “acknowledge[d] that refiners might have concluded in good faith that recoupment on a proportional basis was permitted as a result of possibly ambiguous language in § 212.-83(d) and certain information disseminated by FEA”. Id. at 33283.
Information relating to a survey conducted by the FEA in connection with the proposed “class exception”, which appeared in an article in the Wall Street Journal, resulted in a critical response from the Honorable John D. Dingell, Chairman of the Subcommittee on Energy and Power of the House Committee on Interstate and Foreign Commerce. Following further criticism from Congressman Dingell and discussions between the Congressman and FEA Administrator Zarb and his subordinates,15 the FEA changed its position and announced that it would not grant the proposed class exception, but would require each refiner to “establish . . . that it is subject to a serious hardship or gross inequity as a result of the FEA regulatory requirement”.16 The FEA’s later decision that it would not review the correctness of its interpretation of the regulations in the exception proceedings led to the filing of these lawsuits.
IV. ISSUES ON APPEAL
The issues presented by the respective parties may be summarized as follows:
(1) Should the suits have been dismissed for want of justiciability?
(2) Did the applicable regulations require recovery of all increased product costs before the recovery of any increased non-*1039product costs and preclude the use of (a) the proportional method of recovery and (b) the recovery of increased non-product costs first?
(3) Did the district courts accord proper deference to the FEA’s interpretation of its own regulations?
(4) Did the FEA comply with the procedural requirements of the Administrative Procedure Act and Federal Energy Administration Act in promulgating (a) its cost recovery rule and (b) the non-product cost banking prohibition?
(5) Could the rule requiring the recovery of non-product cost increases last properly be given retroactive effect?
(6) Did the district courts resolve issues of fact in deciding the cases on cross-motions for summary judgment?
V. JUSTICIABILITY
FEA first contends, as it did in the district courts, that the cases were not “ripe” for judicial review under standards set forth in Abbott Laboratories v. Gardner, 387 U.S. 136, 87 S.Ct. 1507, 18 L.Ed.2d 681 (1967); Toilet Goods Association, Inc. v. Gardner, 387 U.S. 158, 87 S.Ct. 1520, 18 L.Ed.2d 697 (1967); and Gardner v. Toilet Goods Association, Inc., 387 U.S. 167, 87 S.Ct. 1526, 18 L.Ed.2d 704 (1967).
The basic rationale of the ripeness doctrine is “to prevent the courts, through avoidance of premature adjudication, from entangling themselves in abstract disagreements over administrative policies, and also to protect the agencies from judicial interference until an administrative decision has been formalized and its effects felt in a concrete way by the challenging parties”. Abbott Laboratories, 387 U.S. at 148^49, 87 S.Ct. at 1515. In deciding whether a case is “ripe” for judicial resolution it is necessary to determine (1) the fitness of the issues for judicial decision and (2) the hardship to the parties of withholding court consideration. Id.
An evaluation of the fitness aspect involves the following factors: whether the agency action was final or whether further administrative proceedings were contemplated; whether the issue presents a purely legal question; and whether judicial determination of the issue “is likely to stand on a much surer footing in the context of a specific application of [the] regulation .....” Toilet Goods Association v. Gardner, 387 U.S. at 162-64, 87 S.Ct. at 1524. In evaluating the hardship aspect, a court must consider whether the regulations have an immediate and direct impact upon the plaintiffs’ conduct of their affairs, with serious penalties attached to non-compliance with the regulations. Abbott Laboratories, 387 U.S. at 153, 87 S.Ct. 1507.
Both district courts, following a careful analysis of all of the factors set forth in Abbott Laboratories and both Toilet Goods cases, concluded that the questions now before this court, were ripe for judicial review. See Standard Oil I, 440 F.Supp. at 357-370; Phillips Pet. I, 435 F.Supp. at 1245-48. We agree. The questions presented are purely legal: (1) a challenge to the FEA’s interpretation of the regulations, and (2) a challenge to the procedures followed by the FEA in adopting certain of the regulations. Although the FEA did institute exception proceedings, it also made clear it would not allow refiners to challenge the agency’s interpretation of the regulations at these proceedings. We find this insistence on the NPCI Last sequency of recovery rule to be a “final” decision. Unless this court takes jurisdiction, these legal questions will be left unadjudicated while the regulations will still apply with full force and effect. These issues are fit for judicial decision.
The test of hardship is also satisfied. The district courts identified these adverse consequences which would result in hardship to the refiners: (1) the impact of the FEA’s interpretation on the refiners’ budgeting and planning functions; (2) the exposure to multiple private, civil enforcement court litigation; (3) the burdensome expense and discovery strategy which the exception proceedings would entail; and (4) the damage to the refiners’ reputations and good will *1040stemming from the FEA’s public announcement of its interpretation and the consequent conclusion by the public of an overcharge by the refiners. See Standard Oil I, 440 F.Supp. at 365-70; Phillips Pet. I, 435 F.Supp. at 1247-48.
We conclude that these cases are ripe for judicial resolution on the legal questions which the district courts decided.
VI. INTERPRETATION OF THE REGULATIONS
The FEA concedes that the regulations in effect during the relevant period did not explicitly impose a non-product cost increase last rule governing the sequence of cost recovery. It argues, however, that the base price concept combined with the prohibition on banking non-product cost increases implicitly required such a sequence of recovery rule.
Both the Ohio and the Delaware courts concluded that for the courts to uphold the FEA’s interpretation, the FEA must establish two points: (1) that the base price was “fixed”, that is, for a particular product and class of purchaser in a given month the base price was equal to the May 15, 1973 selling price plus all product cost increases allocated to that product; and (2) that the base price rule established a sequence for determining prices which also governed the recovery of costs from revenues at the end of the month. Standard Oil II, 453 F.Supp. at 222; Phillips Pet. II, 449 F.Supp. at 774. The Ohio court held that the base price formulation was flexible in that it did not require refiners to include all available product cost increases in their base prices before they could apply increased non-product costs over their base price. 453 F.Supp. at 228.17 The Delaware court concluded that regardless of whether the base price was fixed or flexible, the base price rule did not require the use of a particular sequence of cost recovery. 449 F.Supp. at 774. We conclude that the regulations on their face are ambiguous. While the FEA’s interpretation of the regulations as requiring a NPCI Last method of cost recovery is a reasonable interpretation, in our opinion it is not compelled and the district courts could properly conclude that the NPCI Last method was not required.
(a) Base Price Rule
When the Subpart L regulations governing the pricing of petroleum and petroleum products were adopted by the CLC in August, 1973, the base price concept formed the key to the pricing regulations. At the time base price for the sale of covered products 18 was defined as the May 15, 1973 selling price plus adjustments for the increased costs of imports and domestic crude petroleum. § 150.358(g); 38 Fed.Reg. 22536, 22541 (August 22, 1973). Refiners could pass through these increased costs automatically each month without regard to any of the restraints imposed on the economy’s manufacturing sector by the Phase IV general regulations.19 Refiner price increases to recover cost increases unrelated to increased costs of imports and domestic *1041crude petroleum remained subject to the anti-inflationary and cost absorption rules of the Phase IV regulations applicable to the general manufacturing sector of the economy; that is, these costs could only be passed through after refiners pre-notified the CLC of their intent to increase prices, and then only as a percentage price increase “in excess of the base price”.20
During the relevant period a refiner could not charge a price for a covered product in excess of the base price established for that product except as specified in the regulations. § 212.82(a); 39 Fed.Reg. 42368, 42369 (December 5, 1974). Base price for the sale of a covered product was the “weighted average price ... on May 15, 1973, plus increased product costs . measured pursuant to the provisions of § 212.83”. § 212.82(b); 39 Fed.Reg. at 42369. Section 212.83, entitled “Allocation of refiner’s increased costs”, prescribed the requirements governing the inclusion of a refiner’s increased product cost in the computation of base prices and the inclusion of increased non-product costs in the computation of allowable prices in excess of base prices. Section 212.83 provided that in computing the base price for the sale of a special product “a refiner may increase its May 15, 1973 selling prices . . . once each calendar month ... by an amount to reflect the increased product costs attributable to the sales of that special product . . . provided that the amount of increased costs used in computing a base price is calculated by use of the general formula set forth in paragraph (c)(2)(i) of this section”. § 212.83(e)(l)(i); 39 Fed.Reg. at 42370. “Increased product costs” was defined as “the sum of (1) the difference between the total cost of crude petroleum during the month of measurement and the total cost of crude petroleum during the month of May, 1973 plus (2) the difference between the total cost of petroleum product during the month of measurement and the total cost of petroleum product during the month of May, 1973.” § 212.82(b); 39 Fed.Reg. at 42369.21 The dollar increase of increased product costs which could be added to the May 15, 1973 selling price for special products was represented in the specified formula by the figure “d¡u”. This was defined as “[t]he dollar increase that may be applied to the May 15, 1973 selling price of the special product . . . to compute the base price. . . . ” § 212.83(c)(2)(i); 39 Fed.Reg. at 42370.22
The FEA contends that these regulations compel the conclusion that the base price was a “fixed” rather than a “flexible” concept. Because § 212.82(b) defined “increased product costs” to mean the “total cost” of crude petroleum and petroleum product, the FEA argues that the refiners were required to add all available product cost increases to the May 15, 1973 price to determine a base price. The amount of increased product cost to be added to the May 15, 1973 selling price had to be calculated according to a specified formula, that this formula “yielded one answer”, represented by the figure “diu”, and that this amount was “the total amount of PCI available for recovery . . . .” (DOE Br. at 22) The FEA concludes that the price calculated by adding the “d¡u” dollar increment to the May 15, 1973 selling price of the relevant product was the base price of that product. Only after the calculation of base price was arrived at in this manner could a refiner add into the pricing equation non-*1042product cost increases to obtain a price in excess of base price.
The Ohio plaintiffs argued, and the Ohio court found, that the “base price” was a flexible rather than a fixed concept. The formulas established the maximum allowable base price rather than the sole base price. In arriving at this conclusion the court focused on the regulations as adopted by the CLC. The allocation of increased costs provision, § 150.356(c)(1),23 provided that a refiner “may increase” its May 15, 1973 selling prices to reflect increased costs of imports and domestic crude petroleum provided the amount of increase included was calculated according to a specified formula. In addition, § 150.356(c)(1) was amended in September, 1973, to incorporate a rule for banking unrecouped increased costs. This rule provided that if any firm “establishes a base price for any covered product . . . which does not include the entire amount of increased costs calculated pursuant to [the required] formula ., the unused portion may be added to the May 15, 1973, selling price to compute” the base price for a subsequent month.24 Because these sections were both couched in terms of “may”, the Ohio court concluded that the refiners were vested with discretion “to either pass through increased product costs by adding them to base prices under § 150.356(c) or to bank increased product costs and pass them through at some future date, under § 150.-356(c)(1), when the refiners ‘may’ deem it appropriate to do so”. Standard Oil II, 453 F.Supp. at 225.
*1043The FEA counters that the Ohio court placed too heavy an emphasis on the use of the word “may”, that its reliance upon the wording of the provision for banking product cost increases was misplaced, and that the Ohio court did not give proper deference to the contemporaneous construction given the base price regulations by the CLC and its successors. While the FEA’s arguments have some merit, we do not think that the regulations were so clear as to compel a “fixed” base price definition.
First, the FEA asserts that while the word “may” in the regulations “clearly gave refiners discretion to increase their May 15, 1973 selling prices by adding available PCI, it should have been clear to all who operated under these regulations that the word ‘may’ was not also intended to grant refiners discretion to vary their base prices by including therein less PCI than the value given for ‘d¡u’ or ‘Du¡’ in the formulas”. (DOE Br. at 23) (Emphasis in original.) How this “should have been clear” to refiners the FEA itself does not make clear. It asserts that if refiners could vary the amounts of PCI used in their base price calculations, “there would have been little purpose in having such formulas”. It ignores the fact, however, that the formulas still would have had a purpose — to set the maximum allowable price that refiners could charge.25
We recognized that the interpretation of the word “may” is not constant: in one instance it may be discretionary, while in another it may be mandatory. One must look to the background circumstances and context in which the word is used and the intention of the administrative body which used it to determine its meaning in a particular instance. United States v. Reeb, 433 F.2d 381, 383 (9 Cir. 1970); Wilshire Oil Co. v. Costello, 348 F.2d 241, 243 (9 Cir. 1965). But here we find no circumstances which compel the conclusion that the FEA intended its use of the word “may” to be mandatory. Even though refiners were allowed discretion in adding available PCI to the May 15, 1973 selling prices to determine a base price, the pricing formulas still had a purpose: to establish the maximum price refiners were allowed to charge.26
Second, the FEA contends that the banking regulation was not intended to alter the requirement to include all available PCI in base prices. The FEA argues that the banking provision as initially adopted 27 was *1044intended to provide a mechanism for recovering PCI in a subsequent month when a refiner’s estimate of its current month’s sales volume proved too low. It asserts the language in the first banking provision was “merely descriptive of this failure of a base price predicated upon an estimated sales volume to pass through the entire amount of PCI available for recovery”. (DOE Br. at 25.) (Emphasis added.) It is apparent, however, from an examination of the language of this provision that this construction is not required.
As initially adopted the regulation provided: “If in any month ... a firm establishes a base price for any covered product . . . which does not include the entire amount of increased costs calculated pursuant to the [required] formula . , the unused portion may be added to the May 15, 1973, selling price to compute the respective base price for a subsequent month.” § 150.356(c)(1); 38 Fed.Reg. 25686, 25688 (September 14, 1973). Under the regulation a refiner established a base price at the beginning of the current month. It would not know until the end of the month whether the sales volume it estimated at the beginning of the month (and upon which it based its selling price) would result in an under-recovery of PCI. But the regulation allowed a refiner which had not used all of its increased product costs in establishing a base price to include the unused portion in the following month’s calculation. The “unused portion” refers to that portion unused in establishing a base price, a computation which had to be made before a refiner knew if it had an under-recovery of PCI. Thus this regulation clearly contemplated a situation in which a refiner within its discretion did not include all available PCI in computing its base price.28
Third, the FEA points to its decision in the Gulf Oil Corp. decision as evidence that it contemporaneously construed the base price rule to be a “fixed” concept. The appeal in Gulf Oil arose out of a condition that the FEA attached to its approval of a July 11, 1974, pre-notification request by Gulf to charge a price for gasoline in excess of base price.29 Thomas Músico, an FEA official in the pre-notification department of the FEA, interpreted this condition 30 as requiring refiners to recover all available PCI before they could recover any NPCI. Gulf challenged this interpretation, alleging that the regulations imposed no such sequence of recovery. In its Decision and Order of November 25, 1974 the FEA concluded that Gulf’s contention was without *1045merit (R. 2715, 2716).31 The FEA reasoned that allowable, pre-notified non-product cost increases were to be applied to base prices and “base prices by definition include[d] increased product costs . . ..” Consequently, “Gulf could not use the price increase authorized by the order to increase the price of any covered product which was then being sold below base price levels.” (R. 2219).
The FEA could only have reached this conclusion by interpreting the base price to be “fixed”. Its position here, however, is undercut by the fact that the FEA did not include this condition in all of the pre-noti-fication requests submitted to it during the pre-notification period. Of the twenty pre-notification requests submitted to the FEA between September, 1973, and October, 1974, the FEA inserted this condition in only half of them. See Affidavit of Thomas F. Músico, 1i 6 (September 14, 1977) (R. 2537, 2539-40). Notice of this condition was never published in the Federal Register. In addition the FEA allowed some refiners to pre-notify and pass through at least some non-product costs before fully exhausting all of their product costs, banked and unbanked.32 This action does not jibe with a “fixed” base price interpretation which would require base price to include all available PCI in the base price computation. We, therefore, conclude that the FEA’s interpretation of the base price rule was inconsistent with some of its actions and that a “fixed” base price interpretation was not compelled.33
(b) Sequency of Recovery
The Delaware court concluded that neither the base price rule, the product cost banking provision, nor the prohibition against non-product cost increases required the use of a particular sequence of cost recovery. Phillips Pet. II. 449 F.Supp. at 774, 777, 778.
1. Base Price and Profit Margin Rules
The FEA argues that permitting refiners to recover their increased product and non-product costs in a sequence other than NPCI Last would vitiate both the “base priee/price in excess of base price” and the profit margin limitation rules.
The Delaware court found “absolutely no indication in the regulations that the base price concept applie[d] to anything other than pricing”. Phillips Pet. II, 449 F.Supp. *1046at 775. The intent of § 212.82(c),34 part of the “Price rule” regulation, was to keep prices down. It explicitly prohibited a refiner from implementing a price increase for any reason other than to recoup increased costs, but it did “not address the issue of cost recovery”. Id. Neither did the FEA contemporaneously construe the regulations to require a particular sequence of cost recovery. Id. at 783.
The FEA claims that the Delaware court’s ruling is “squarely contradicted” by the FEA’s decision in the Gulf Oil Corp. Appeal. In the Gulf Appeal, Gulf claimed that the petroleum pricing regulations did not require recovery of all available product cost increases prior to recovering any pre-notified increased non-product costs. (R. 2699, 2700-01) This question was raised in the context of a pre-notification request by Gulf to charge a price in excess of base price. Upon examination of that decision the Delaware court concluded that the decision “never explicitly addressed the method of cost recovery”. Id. at 776. We agree. The Gulf decision held that “under the provisions of Section 212.82 . . . , Gulf could not use the price increase authorized by the Order to increase the price of any covered product which was then being sold below base price levels”. (R. 2715, 2719) (Emphasis added.) This clearly speaks to what prices the FEA would allow Gulf to charge and not to the sequence of cost recovery. Significantly, top officials at Gulf also concluded that the FEA’s decision did not resolve the question raised in its appeal, and thought this and other circumstances confirmed their belief that no order of recovery was required. Affidavit of L. G. Armel, Is 14-27 (September 27, 1977) (pp. 2997, 3001-10).
The FEA contends that the Delaware court’s holding “would open a judge-made loophope in the profit margin limitation”. (DOE Br. at 32). The profit margin limitation provided:
A refiner which charges a price for any item in excess of the base price for that item in any fiscal year may not for the fiscal year in which the price in excess of the base price is charged, exceed its base period profit margin, as defined in § 212.-31.
§ 212.82(d); 39 Fed.Reg. 42368, 42369 (December 5, 1974)35 The FEA argues that if refiners were not required to charge a price in excess of base price as a precondition to recovery of NPCI, the refiners could “finagle” with the profit margin limitation rule.
FEA recognition of this ability to finagle with the profit margin limitation was brought to light in a meeting on October 22, 1975, between representatives of FEA’s Office of General Counsel and Office of Compliance. This meeting resulted in a memorandum from Gordon Harvey, Director of the Office of Compliance Program Development, to J. Peter Luedtke, the Deputy General Counsel for Pricing. Although Harvey recognized that the pro-rata method of recovery, which many refiners used, permitted some finagling with the profit margin limitation rule, he concluded that that method of recovery was a permissible interpretation of the regulations36 and that rule-*1047making would be required to clarify the cost recovery sequence if the FEA intended to require a PCI first method of recovery. Memorandum, Non Product Cost Recoveries (November 12, 1975) (R. 3593).
This conclusion of the FEA’s Director of Compliance is hardly consistent with its contention that the regulations compelled an NPCI Last sequence of recovery rule. The profit margin limitation itself does not expressly refer to methods of cost recovery. It simply places a limitation upon the amount of NPCI a refiner may include in calculating the prices it may charge. If the regulations permitted methods of recovery which would result in manipulation of the profit margin limitation, it was incumbent upon the FEA to articulate and publish a sequence of recovery rule. This is what Harvey recognized in proposing formal rulemaking.
2. The Product Cost Banking Provision Subsection (e)(1) of § 212.83, entitled “Carryover of Costs”, provided that if in
any month the prices charged resulted in the recoupment of less than the entire amount of increased product costs, the amount not recouped “may be added to the May 15, 1973 selling prices for that special product for a subsequent month”. 39 Fed. Reg. 42368, 42372 (December 5, 1974).37 The FEA implemented this provision through Form FEO-96, a “Monthly Cost Allocation Report”. (See R. 2736 and n. 26, supra). Part IV of that form dealt with the price adjustment data pertaining specifically to gasoline. The instructions to this part directed refiners to compute their banks for product cost increases allocable to gasoline by subtracting “the total amount of revenues received, during the Period of Measurement, in excess of applicable May 15, 1973 selling prices of gasoline” from “the total number of dollars of increased costs which were attributable to gasoline and available for recovery in the Period of Measurement”. (R. 2934, 2939). The FEA argues that § 212.83(e), Form FEO-96, and *1048the instructions thereto, “clearly prohibited a refiner from computing its product cost banks by subtracting an amount less than the ‘total revenues’ recouped through price increases from that month’s ‘entire amount’ of available PCI”. (DOE Br. at 34).
On the face of it the language of the instructions tends to support the FEA’s argument that a refiner’s monthly revenues in excess of the May 15, 1973 selling price had to be subtracted from all PCI available for recovery in that month. In turn this lends credence to the FEA’s interpretation that the PCI banking provision required a NPCI Last method of cost recovery. What the FEA ignores, however, are the circumstances under which the refiners operated during the relevant period, which confused the procedure they were to follow in completing this form.
During the period prior to January 1, 1975, refiners were required to report their increased product costs and increased non-product costs on two separate forms — Form FEO — 96 (for product costs)38 and Form CLC — 22 (for non-product costs).39 In September, 1974, the FEA recognized that the procedure for pre-notifying non-product cost increases had not proved to be “particularly well-suited” for application to refiners. It proposed to drop the pre-notification requirement and to allow refiners to pass through allowable non-product cost increases. To implement this a new Form FEA-96 was to be issued to replace the Forms FEA-96 [FEO-96] and CLC-22.40 After the requirement to pre-notify proposed non-product cost increases was deleted from the regulations,41 Form CLC — 22 became obsolete. Significantly, however, the new form to replace Forms FEO-96 and CLC-22, which the FEA first mentioned in September, 1974, was not finally adopted until April, 1976, after the relevant period. Thus Form FEO — 96 was used throughout the relevant period and was never amended to specify a method for computing the amount of non-product cost increases to be recovered each month. In light of these circumstances we find FEO-96 to be an incomplete picture of all of the computations required of the refiners.42
In its opinion the Delaware court posited a hypothetical situation demonstrating the FEA’s interpretation of the PCI banking provision. Under the FEA’s interpretation of this provision a refiner charging a price above base price was required to apply all revenues received in excess of the May 15, 1973 selling price toward recovery of available product cost increases before applying any portion of those revenues toward recovery of non-product cost increases. A refiner, therefore, that “(1) had a total of $1,000,000 in PCI available for recovery in the current month, (2) had $200,000 in approved pre-notified NPCI, (3) had established a selling price at the beginning of the current month which included all those costs, and (4) as a result, had received revenues in excess of the May 15, 1973 selling price totaling $1,100,000, would recover all $1,000,000 of its PCI and $100,000 . of the available NPCI”. The district court pointed out that contrary to the FEA’s interpretation this hypothetical situation would not result in $100,000 of recovered NPCI. The computations required by section 212.-83(d)43 and the instructions to Form FEO-96 would produce a PCI over-recovery of $100,000 which would have to be subtracted from the May 15, 1973 selling price to compute a base price for the subsequent month. Phillips Pet. II, 449 F.Supp. at 777.
*1049The FEA counters that the language of the banking regulation must be reconciled with the language of the price rule permitting NPCI recovery when a refiner charges a price in excess of the base price. It contends that “[t]hese two provisions are easily reconciled by allowing a refiner to offset what otherwise would be a PCI over-recovery with available NPCI when the refiner recovered more than its base price”. (DOE Br. at 35). It may well be that this construction of the two regulations would have been permissible had the FEA articulated a cost recovery rule. Clearly, however, it is not compelled. As appellees point out, the FEA’s “offset” required a departure from Form FEO-96 and controverts its own banking regulations. Section 212.-83(e)(1)44 expressly required PCI over recoveries to be deducted from the May 15, 1973 selling prices in the subsequent month; it did not permit refiners to offset the over-recoveries against NPCI available for recovery in the current month. We agree with the district court that “the agency’s silence on the method of cost recovery is ambiguous” and “the banking provision and Form FEO-96 do not require the use of the NPCI Last method”. 449 F.Supp. at 777.
3. Prohibition Against Banking Non-product Cost Increases
The FEA argues that the refiners should have recognized the existence of a NPCI Last method of cost recovery from the ban against banking non-product cost increases. It contends that this ban, “if not alone sufficient to compel refiners to recover NPCI Last, was [at the least] an integral part of a regulatory scheme carefully designed to permit refiners to pass through NPCI only on top of base prices (which included all PCI) and to force refiners to absorb these cost increases when current market conditions would not support prices in excess of base prices”. (DOE Br. at 37).
The ban on NPCI banking stated in pertinent part:
Increased non-product costs calculated pursuant to § 212.87 for the month of measurement which are not recouped in the current month (which is the month immediately succeeding that month of measurement) may not be carried forward for use in computing allowable prices in excess of base prices in any subsequent month.
§ 212.83(e)(4); 39 Fed.Reg. 42368, 42372 (December 5, 1974). Although this ban was not part of the petroleum pricing regulations as initially adopted by the CLC (it was promulgated as part of the December 1, 1974 rulemaking), the FEA argues that it was implicit in the regulations from the first. When the FEA did expressly adopt the ban, it did so without complying with the notice and comment provisions of either the Administrative Procedure Act or the Federal Energy Administration Act. Accordingly, as hereinafter set forth,45 we hold the rule invalid. Even if we were to find that the regulation was properly promulgated, we would still conclude that it does not compel a NPCI Last sequence of recovery.
The FEA argues that the ban was part of a “carefully designed” scheme of regulations governing cost recovery. We cannot agree. In his deposition David G. Wilson, Deputy General Counsel of the FEA and one who was primarily responsible for drafting many of the regulations, testified that when the initial petroleum pricing regulations were adopted, alternative methods of cost recovery were not considered.46 Nothing surrounding the adoption of the express ban on banking NPCI leads to a different conclusion.
Prior to December, 1974, the general pricing formulas provided different methods by which refiners calculated allowable price increases for product cost increases and for non-product cost increases. The non-product cost calculation was based upon the “rate of increase” of non-product costs in *1050the month preceding pre-notification over the same costs incurred during the historical base period. The product cost calculation was based upon a “dollar-cost pool” in which the allowable product cost increase was determined by the dollar amount of cost increases incurred rather than a rate or percentage of increase.47 While the dollar amount that could be used to justify a product cost increase had to be recalculated on a monthly basis, the dollar amount used to justify a non-product cost increase (thus resulting in a price in excess of base price) could be used in recovering increased non-product costs over a 12 month period48 subject to the limitation that the non-product cost increases which supported that price in excess of base price continued to be incurred.49 As the Ohio court recognized, the provision allowing recovery of non-product costs to be spread out over a 12 month period did not require recovery of non-product costs to match actual cost increases on a month-to-month basis. Standard Oil II, 453 F.Supp. at 234. Thus a refiner had some discretion in deciding how to allocate these cost recoveries over the 12 month period. See Wilson Depo. 37-40 (R. 1073, 1110-1113).50
In addition, the rulemaking proceeding which led to the adoption of the NPCI banking ban did not indicate that the ban would force refiners to absorb substantial amounts of non-product cost increases. Yet this is precisely what would have occurred with respect to the cost of refinery fuel if the FEA in fact intended the NPCI banking ban to impose a NPCI Last sequence of cost recovery. Refinery fuel, which was fuel used to operate a refinery, amounted to a substantial cost item for refiners. Prior to December 1974 many refiners had treated refinery fuel as a product cost rather than a non-product cost. As such, refiners were able to pass through this cost item in the base price of their products, without the pre-notification and profit margin limitation requirements applying to it.
*1051Two of the modifications, proposed in the September 6, 1974, notice of proposed rule-making, affected the treatment of refinery fuels. First, the FEA intended to delete the requirement that refiners pre-notify their proposed non-product cost price increases. It advocated a single method of calculating cost pass through for both product and non-product cost increases, with non-product costs remaining subject to the profit margin limitation. Second, refinery fuel costs were to be treated by all refiners as a non-product cost.
A NPCI Last sequence of cost recovery would have forced many refiners to absorb their refinery fuel costs rather than to pass through and recover them. Although the FEA recognized refining fuel as an “important cost item” and sought “equitable treatment” for the refiners, it gave no indication that classification of refinery fuel as a non-product cost could have a substantial effect on refiners. Instead it assured refiners that the proposed changes with respect to the passthrough of non-product cost increases would “permit refiners generally to pass through such increased costs of refinery fuels, subject to a profit margin limitation”. Preamble, 39 Fed.Reg. 32718, 32725 (September 10, 1974). In marked contrast to its silence regarding the effect of treating refinery fuel as a non-product cost is what the FEA said concerning its redefinition of “non-product costs”. While recognizing that the new definition would “limit the increased non-product costs which a refiner may pass through in price increases”, it concluded that the change would have “little overall impact”, as the omitted costs comprised a relatively small proportion of most refiners’ overall costs. Preamble, 39 Fed.Reg. 32721-22. Still, if any significant cost item had been omitted from the cost pass through provisions, the FEA was willing to consider amending the regulations. Id. These comments plus the failure to consider the impact of making refinery fuel a non-product cost are hardly consistent with a conclusion that the FEA intended the ban on NPCI banking to impose a NPCI Last sequence of recovery.
4. EPAA Objectives
Section 4(b)(1) of the Emergency Petroleum Allocation Act of 1973, 15 U.S.C. § 753(b)(1) (1976), sets out nine objectives for which the petroleum pricing regulations were to provide to the maximum extent possible. Among them is “equitable distribution of [petroleum products] at equitable prices . . . .” 15 U.S.C. § 753(b)(1)(F) (1976). The FEA argues that the decisions of the Ohio and Delaware courts allowed the refiners “to contravene the spirit if not the letter of [§ 753(b)(1)(F)]”. (DOE Br. at 38). It contends that the district courts’ opinions negate the cost absorption principles of the regulations, thereby allowing refiners to charge prices which are not equitable.
The district courts’ opinions do not, however, totally negate the cost absorption principles of the regulations. Additionally a balancing among the nine objectives of the EPAA is required. Other objectives require “preservation of an economically sound and competitive petroleum industry”; “economic efficiency”; and “minimization of economic distortion, inflexibility, and unnecessary interference with market mechanisms”. 15 U.S.C. §§ 753(b)(1)(D), (H) and (I) (1976). As the FEA itself realized when it allowed comment on its express adoption of the NPCI Last sequence of cost recovery, that sequence would create great economic distortion through rapid price fluctuations. Consequently it abandoned the NPCI Last method for the period subsequent to February 1, 1976. We think the district courts’ opinions reflect the EPAA objectives and do not find them contrary to the intent of Congress in adopting the EPAA.
5. Conclusion re: Sequence of Recovery Rule
In both the district courts and in this court the FEA has constantly maintained that the petroleum pricing regulations established rules governing a well identified sequence by which refiners were to deem their costs recovered. It has argued that the two district courts’ opinions rendered *1052meaningless the anti-inflationary and forced cost absorption features of those regulations, leaving them without purpose. We think this incorrect.
The pricing regulations did establish a procedure for setting a maximum price refiners could charge. Through the provision for banking product cost increases, refiners could voluntarily delay price increases, thereby smoothing out price increases. As the FEA came to realize from the events that occurred after it expressly adopted a product cost increase first sequence of cost recovery in February, 1976, such a sequence of recovery combined with a ban on banking non-product cost increases was not anti-inflationary. It gave refiners incentive to accelerate recovery of all available cost increases in order to avoid having to absorb non-product cost increases. At that point the FEA concluded that the NPCI Last recovery rule, first expressly stated on February 1, 1976, should be repealed to that date.51
In conclusion we reiterate that the petroleum pricing regulations in effect during the relevant period did not expressly impose a sequence of recovery rule. We agree with the district courts that the regulations were ambiguous. There was confusion and uncertainty among both FEA officials and the refiners. While the regulations could reasonably be interpreted as imposing a NPCI Last sequence of recovery, in our opinion this interpretation was not compelled. The regulations could also reasonably have been construed as imposing no particular sequence of recovery.
VII. DEFERENCE TO THE FEA’S CONTEMPORANEOUS CONSTRUCTION
The FEA claims that throughout the relevant period it consistently construed the applicable regulations as requiring refiners to follow a NPCI Last sequence of cost recovery. Although it recognizes that lower level FEA officials issued contrary advice during this period, it maintains this advice was “unauthorized and erroneous” and should be ignored by this court.52 It contends that agency policy makers David G. Wilson and Gorman Smith did not become aware of the need to clarify the agency’s position on the sequence of cost recovery until December, 1975. As noted supra, the agency then included a statement in the preamble to its February 1, 1976, rulemak-ing that the regulations had always required a NPCI Last sequence of cost recovery. The FEA concludes that this court must defer to this official interpretation of the regulations as they existed during the relevant period.
In resolving this issue it is first necessary to examine the various interpretations by FEA officials during the relevant period. These are set out in detail in the district courts’ opinions.53 We shall not repeat the findings of the two district courts, except to emphasize a few of the publications and memoranda which are most significant in determining whether this court should defer to the Agency’s present interpretation.
(a) Contemporaneous Construction
After the December, 1974, amendments to the petroleum pricing regulations became effective, Donald Clyman, Acting Chief of the FEA’s Refinery Audit and Review Program (RARP) directed Andrew Drance, a case analyst on his staff, to prepare an explanation of the December amendments to its field auditors. The result, Regulation Change Notice 3-1975-1 (Change Notice)54 was distributed to the field auditors in January, 1975, for insertion in their Compliance Audit Review Division (CARD) Audit Handbooks. The Change Notice indicated *1053that product and non-product cost increases should be recovered proportionately.
Although Clyman and Drance recognized that § 212.83(d)55 expressly applied only to the allocation of NPCI among the various product categories for purposes of computing allowable prices in excess of base prices, they believed the intent of that subsection was to establish a proportional mechanism for the recovery of increased product and non-product costs. They arrived at this conclusion because they perceived a conflict between § 212.83(d) and § 212.87(b), which they also viewed as prescribing a method for allocating NPCI among the various product categories for purposes of computing prices in excess of base prices. Additionally both Clyman and Drance believed that a proportional method of recovery “was reasonable as a matter of regulatory policy”.56 Thus, they included within the Change Notice an example problem demonstrating how to compute recoveries by a proportional method when both PCI and NPCI were used in computing price adjustments to the May 15, 1973, selling prices.57
On March 5, 1975, Andrew Drance and Fred Stuckwisch from the Office of Compliance, W. Mayo Lee from the Office of General Counsel, and others met to discuss revisions to Form FEO-96 necessitated by the December, 1974, amendments to the regulations. The Office of Compliance, which had previously issued the Change Notice and supplemental worksheets directing its field auditors to use the proportional method of cost recovery, advocated interpreting the regulations to require that particular sequence. Mayo Lee and others from the Office of General Counsel favored a NPCI Last interpretation.58
On March 26, 1975, Laura Kuitunen and Mayo Lee, both from the Office of General Counsel, drafted a memorandum to Don Clyman commenting on certain aspects of the proposed form. They stated that “[t]he pro-rata recovery of product vs. non-prod*1054uct costs is not set forth in the regulations”. Memorandum, p. 2 (R. 3262-63). Consequently, on March 31, 1975, the group redrafting Form FEO-96 decided to strike all language from the proposed form concerning recovery of cost increases. A memorandum explained the group’s reasoning: “There is nothing in the regulations that requires such a pro-rata application and it was decided that until something is made clear on this issue, the form can reflect only what is in the regulations.” (R. 3264). A month later the FEA published a proposed revision of Form FEO-96, entitled Form FEA-P110-M-1, in the Federal Register. 40 Fed.Reg. 19120 (May 1, 1975). Schedule E of that form provided for a proportional allocation of non-product cost increases among different refined products, but it said nothing about the method in which non-product costs were to be recovered.59 The FEA received comments on the proposed form, published a revised proposed Form FEA-P110-M-1 in October, 1975, but did not finally adopt a new form until April, 1976, after it had abandoned the NPCI Last sequence of cost recovery.
In April, 1975, Andrew Drance sent a memorandum to Peter Luedtke, the Assistant General Counsel for Pricing, in an attempt to obtain clarification from the Office of General Counsel as to the correct method of recovery.60 The memorandum recognized cost recovery as an “important item” and noted that refiners had questioned how product and non-product costs were to be recovered when they charged prices which were less than the maximum permissible price in excess of base price. Drance indicated that the Office of Compliance had advised refiners to use the proportional method of recovery and concluded that if this method was proper, it should be “spelled out” by an amendment to the regulations.61
On September 4, 1975, the FEA issued Ruling 1975 — 16.62 This Ruling, which applied to resellers and retailers, interpreted the pricing regulations governing them to require a NPCI Last sequence of cost recovery. Because the regulations governing resellers and retailers were similar to those governing refiners, Ruling 1975-16 presented an apparent conflict with the Change Notice which RARP had issued earlier in the year. A meeting on October 22, 1975, between Drance and members of the Office of General Counsel resulted in the memorandum from Gordon Harvey, Director of the Office of Compliance, to Peter Luedtke, referred to supra, in our discussion of the profit margin limitation. In this memorandum Harvey stated that following the issuance of the December 5, 1974 regulations, containing sub-paragraph 212.83(d), “the general interpretation held by both General Counsel and Compliance at that time was that recoveries of non-product cost increases was to be made on a pro-rata basis with the recovery of increased product costs”. Although recognizing that the proportional method allowed some “finagling” with the profit margin limitation rule, Harvey maintained that § 212.83(d) could reasonably be interpreted to permit proportional recoveries of product and non-product cost increases even when the selling price was below base price. He concluded that a rulemak-ing would be required to impose a NPCI Last sequence of cost recovery.63
*1055That refiners informed high level FEA officials they were recovering their costs in other than a NPCI Last sequence is also clear from the record. As early as February, 1975, Shell Oil Company wrote a letter to Gorman Smith, the FEA’s Assistant Administrator of Regulatory Programs, in which it stated that it understood the intent of § 212.83(d) was to allow proportionate recovery of non-produet cost increases.64 On April 29, 1975, Mobil Oil Company submitted a formal request to the FEA’s General Counsel for an interpretation of the non-product cost allocation provisions in § 212.87 and of the banking regulations in § 212.83. Mobil stated that it was banking unrecovered non-product cost increases and claimed that “while unrecovered non-product costs may not be used to increase base prices in a future month, there is no prohibition in the regulations against using such non-product costs to maintain existing base price levels”.65 And in August, 1975, Amoco Oil Company wrote a letter to Gorman Smith confirming a conversation with him involving Amoco’s having used a proportional method of cost recovery in the first three months of 1975.66
On January 7, 1976, the FEA announced proposed amendments to the petroleum pricing regulations necessitated by Congress’ adoption of the Energy Policy Conservation Act of 1975 (EPCA), Pub.L. 94-163, Title IV, 89 Stat. 941, but it said nothing about the sequence of cost recovery under the regulations.67 Finally, on February 1,1976, the FEA issued a “new” regulation (§ 212.85) which for the first time expressly adopted the NPCI Last sequence of cost recovery,68 stating in the preamble that the cost recovery sequence specified in § 212.85 “is the same as that under the regulation previously in effect”.69 Two months later § 212.85 was repealed retroactively to February 1, 1976.70
After detailing the contemporaneous construction by various officials of the FEA, the Delaware court reached the following conclusion, in which we concur.
The record of contemporaneous construction suggests that confusion existed within the FEA and among refiners regarding the proper method for computing cost recoveries, and that the FEA, which presumably could have eliminated the confusion by interpreting the regulations to require a particular method, did not decide which method was correct until the end of the relevant period.
449 F.Supp. at 792.
(b) Deference to Administrative Interpretation
It is well established that where administrative regulations are ambiguous on their face, the court should look to the construction which the responsible agency has given to them. Bowles v. Seminole Rock & Sand Co., 325 U.S. 410, 413-14, 65 S.Ct. 1215, 1217, 89 L.Ed. 1700, 1702 (1945). The primary rationale behind the doctrine of deference is recognition of administrative expertise developed through implementation and enforcement of statutes and regulations. Consequently, an agency’s interpretation is normally deemed controlling unless it is plainly erroneous or inconsistent with the regulations. Udall v. Tallman, 380 U.S. 1, 16-17, 85 S.Ct. 792, 801, 13 L.Ed.2d 616, 625, reh. denied 380 U.S. 989, 85 S.Ct. 1325, 14 L.Ed.2d 283 (1965).
*1056Deference to an agency’s “interpretation”, however, is not a hard and fast rule. The weight to be given to an administrative interpretation depends upon “the thoroughness evident in its consideration, the validity of its reasoning, its consistency with earlier and later pronouncements, and all of those factors which give it power to persuade, if lacking power to control”. Skidmore v. Swift & Co., 323 U.S. 134, 140, 65 S.Ct. 161, 164, 89 L.Ed. 124, 129 (1944). In Udall the Court deferred to the agency’s interpretation of regulations which had long “been a matter of public record and discussion”. Udall, 380 U.S. at 17-18, 85 S.Ct. at 802, 13 L.Ed.2d at 626. On the other hand, in Davies Wharehouse Co. v. Bowles, 321 U.S. 144, 156, 64 S.Ct. 474, 481, 88 L.Ed. 635, 644 (1944), the Court did not defer to an administrative construction of a controlling statute which was challenged as soon as it was made and had not “seasoned or broadened into a settled administrative practice”.
We do not, of course, question the rule of deference set forth in Udall v. Tallman and related cases. This rule is simply not applicable to the FEA’s interpretation first announced on February 1, 1976, at the end of the relevant period. The FEA relies upon an after the fact interpretation in the preamble to the new rule that, “The order specified in the new Section 212.85 is the same as that under the regulations previously in effect.” Yet, it is undisputed that during the relevant period the only public pronouncements were made by officials of the Office of Compliance and the FEA auditors and were contrary to the interpretation set forth in the February 1, 1976 rule.
The FEA contends that only the FEA’s General Counsel, his staff, and other “high level policy makers” had the authority to issue official interpretations of its regulations. Consequently, it argues, in determining what the agency’s interpretation was this court should ignore the actions of the FEA auditors and other lower level officials during the relevant period. This court held in California Molasses Co. v. California & Hawaiian Sugar Co., 551 F.2d 1230, 1235, 1239 (Em.App.1977), that the interpretation of agents of the IRS, which had been charged with enforcing price controls, were entitled to deference by the courts. We recognize, as the FEA argues; that California Molasses is not precisely in point. We do conclude, however, that the statements by the FEA auditors and other lower level officials are entitled to weight in determining the thoroughness of the FEA’s consideration of its regulations, the validity of its reasoning, and its consistency with earlier and later pronouncements. In any event, we find no basis for according deference to the interpretation of the FEA first announced on February 1, 1976.
VIII. COMPLIANCE WITH RULEMAKING REQUIREMENTS
(a) Provisions of the Administrative Procedure Act
Section 4(b)(3) of the Administrative Procedure Act (APA), 5 U.S.C. § 553(b)(3), requires agencies in rulemaking to publish “either the terms or substance of the proposed rule or a description of the subjects and issues involved”. Section 4(c), 5 U.S.C. § 553(c), requires agencies to “give interested persons an opportunity to participate in the rulemaking through submission of written data, views or arguments . . .” Section 4(b), however, does not apply to “interpretative rules, general statements of policy, or rules of agency organization, procedure, or practice” or “when the agency for good cause finds (and incorporates the finding and a brief statement of reasons therefor in the rules issued) that notice and public procedure thereon are impracticable, unnecessary, or contrary to the public interest”. 5 U.S.C. §§ 553(b)(A) and (B). Section 4 of the APA was incorporated by reference in Section 5(a)(1) of the Emergency Petroleum Allocation Act of 1973, 15 U.S.C. § 754(a)(1). The applicability of these procedural provisions of the APA to the Federal Energy Administration has been considered by this court in a number of cases involving various factual situations. In resolving this difficult and crucial issue *1057under the unusual circumstance of this case, a review of prior decisions is helpful.
In California v. Simon, 504 F.2d 430, 439 (Em.App.1974), cert. denied, 419 U.S. 1021, 95 S.Ct. 496, 42 L.Ed.2d 294 (1974), an action against the administrator of the FEO, we held that “substantial compliance with rule making requirements is essential to the validity of administrative rules”, but we declined to follow “a doctrine of technical absolutism”.71 Following California v. Simon, we held in Nader v. Sawhill, 514 F.2d 1064, 1068-69 (Em.App.1975) that while “good cause” to dispense with procedural requirements was not expressed in that case by the Cost of Living Council, this was a technical violation and “good cause in fact was present”. We warned, however, “that repeated technical noncompliance will not be tolerated” and stressed “categorically that our resolution of the procedural issues” was “founded upon the unique circumstances in which this price increase was formulated”.
In Shell Oil Co. v. FEA, 527 F.2d 1243, 1248 (Em.App.1975), in holding that the FEA may not, pursuant to its authority under the EPAA, regulate the rent charged for real property used in retailing gasoline, we rejected FEA’s “good cause” justification for issuing price regulations without notice and held that the challenged regulations were invalid. Following Shell, in Tasty Baking Co. v. Cost of Living Council, 529 F.2d 1005, 1014-15 (Em.App.1975), we held invalid a regulation promulgated without notice which “defined for the first time profit margin violations”. We noted our concern with agency actions which evidenced “a cavalier disregard of procedural requirements”.72
Our opinion in National Helium Corp. v. FEA, 569 F.2d 1137, 1144-46, (Em.App. 1977),73 reviewed our prior cases and considered in particular the “good cause” and “interpretative rules” exceptions to the notice and comment requirements of the APA. With respect to the latter exception, we summarized the applicable principles as follows:
In construing this exception, rather than rely on a “facile semantic distinction” between interpretative rules and substantive or legislative rules, courts have looked instead to the “basic purpose of [the] statutory requirements [of notice and comment]” in deciding whether the requirements should be imposed. Pharmaceutical Manufacturer’s Association v. Finch, 307 F.Supp. 858, 863 (D.Del.1970). When an agency action has “palpable effects” upon the regulated industry and the public in general, it is necessary to expose that action “to the test of prior examination and comment by the affected parties.” National Motor Freight Traffic Association v. United States, 268 F.Supp. 90, 96 (D.D.C.1976) (three judge court) aff’d per curiam, 393 U.S. 18, 89 S.Ct. 49, 21 L.Ed.2d 19 (1968).
The APA’s rulemaking procedures “were designed to assure fairness and mature consideration of rules of general application”. NLRB v. Wyman-Gordon Co., 394 U.S. 759, 764, 89 S.Ct. 1426, 1429, 22 L.Ed.2d 709 (1969). The prior publica*1058tion and opportunity for comment requirements enable “the agency promulgating the rule to educate itself before establishing rules and procedures which have a substantial impact on those regulated”. Texaco, Inc. v. FPC, 412 F.2d 740, 744 (3 Cir. 1969).74
(b) Provisions of the Federal Energy Administration Act
Sections 7(i)(l)(B) and (C) of the Federal Energy Administration Act, 15 U.S.C. § 766(i)(l)(B) and (C), were applicable to the FEA from the time the FEAA was enacted on May 7, 1974 until they were repealed in 1977 by the Department of Energy Organization Act. Section 7(i)(l)(B) required that notice of any proposed rule or regulation “shall be given by publication” in the Federal Register, and that a “minimum of ten days following such publication shall be provided for opportunity to comment”. Section 7(i)(l)(C) provided that if the rule “is likely to have a substantial impact on the Nation’s economy or large numbers of individuals or businesses, an opportunity for oral presentation of views, data, and arguments shall be afforded. To the maximum extent practicable, such opportunity shall be afforded prior to the issuance of such rule, regulation, or order, but in all cases such opportunity shall be afforded no later than forty-five days after the issuance of any such rule, regulation, or order.”75 As the court recognized in Phillips Pet. II, these requirements of the FEAA are “more rigorous than those provided for in the APA”. 449 F.Supp. at 799-800.
These FEAA notice requirements were construed by this court in Shell Oil Co. v. Federal Energy Administration, 574 F.2d 512, 516 (Em.App.1978), where we held that a regulation which determined prices for unleaded gasoline by reference to those for premium gasoline was invalid for failure to comply with the statutory requirements. After recognizing that where defects in the compliance with procedural requirements do not defeat the rulemaking process courts have been unwilling to invalidate the resulting rule, we stated:
However, the deficiencies in the rulemak-ing procedures in the instant case were more than technical; they defeated the process. Congress recognized that the FEA’s final judgment could only be as good as the information upon which it drew, and prescribed certain procedures— stricter than those of the Administrative Procedure Act (‘APA’) — which would permit it to be educated by the most interested and therefore, hopefully, the most knowledgeable parties. See Texaco v. Federal Power Commission, 412 F.2d 740 (3d Cir. 1969). The agency’s discretion, and thus its latitude for promulgating unwise rules, was to be restrained through this deliberately prescribed process for meaningful comment — a competition of ideas, which replaced the restraint once performed by the marketplace’s competition for profits. Cf. Steward, The Reformation of American Administrative Law, 88 Harv.L.Rev. 1669 (1975). Here, the procedural shortcomings undermined *1059the effectiveness of the educational process which Congress intended the FEAA to provide for agency rulemaking.76
This court considered the notice and comment requirements of the APA and FEAA most recently in Energy Reserves Group et al. v. Department of Energy et al. and related cases, 589 F.2d 1082 (Em.App.1978), with one member of the panel concurring specially, and one member dissenting. We construed a ruling interpreting a regulation and underlying statutory provisions exempting crude petroleum produced by “stripper wells” from allocation and price regulation. We upheld as an “interpretative rule” under the APA exception, 5 U.S.C. § 553(b), a ruling which excluded injection wells from the “well count” for purposes of applying the stripper well exemption.
The majority agreed that the ruling in question had no “substantial impact” on the parties and that the rulemaking requirements of the FEAA set forth in § 7(i) of the Act were not applicable to an interpretative rule.77
(c) Application of the APA and FEAA Rulemaking Requirements to Rules in these cases
We turn now to a consideration of the application of the APA and FEAA rule-making requirements, in the light of the prior decisions of this court, to both the NPCI Last recovery rule and the rule prohibiting the banking of non-product cost increases.
1. NPCI Last Recovery Rule
In holding the NPCI Last recovery rule invalid for failure to comply with both the APA and FEAA requirements, the court in Standard Oil II said:
In adopting the all product costs first theory of increased cost recovery, the FEA and its predecessor agencies never complied with these procedural requirements. The FEA continues to rely on the masquerade that compliance with these requirements was not necessary because it never changed the implicit requirements of the regulations. Yet, this court has found that nowhere prior to the February 1, 1976 rulemaking did the agency ever give notice that it required an all product cost first sequence for the re-coupment of increased product and non-product costs. The agency did not furnish advance notice of the subject matter or the issues presented by such a rule. The agency did not furnish advance notice of the rule itself, or provide the pub-*1060lie with the required opportunity for the presentation of written comments. Nowhere did the FEA furnish the public the required opportunity for oral comment, despite the rule’s enormous impact, and nowhere did the agency publish a statement of basis and purpose for such a rule. 453 F.Supp. at 243.
The court in Phillips Pet. II quoted the foregoing with approval, “concur[ring] fully”. 440 F.Supp. at 801.
It is clear that no express sequence of recovery rule had been adopted by the FEA prior to February 1, 1976. It is likewise clear that no notice of the proposed NPCI Last recovery rule was given prior to its adoption. FEA does not contend otherwise, but on appeal argues that (1) the regulations in effect during 1975 and January, 1976 “compelled the NPCI Last pass-through sequence”, and (2) if FEA’s interpretation was not “compelled”, it “validly set forth its interpretation in accordance with applicable procedure”. In support of this contention, it relies upon the publication of its “proposed interpretation” on February 4 and March 3, 1976, which provided more than 10 days following publication for an opportunity to comment, and afforded an opportunity to make oral presentation of “views, data and argument”. In other words, the FEA relies upon notice of its “interpretation” after the relevant period, to be applied retroactively during that period.78
We have concluded that the FEA’s NPCI Last recovery rule was a reasonable interpretation of existing regulations and did not exceed the FEA’s statutory authority. We agree with the district court in Standard Oil II, that “had the agency issued regulations which actually embodied that interpretation, pursuant to the required statutory procedures mandating prior publication, notice, and an opportunity to comment, such an interpretation would have been valid for the time frame January 1, 1975 through January 31, 1976”. 453 F.Supp. at 245. For the reasons set forth supra, we do not agree with FEA, however, that its interpretation was “compelled” by existing regulations. Nor do we find that its interpretation was “contemporaneous” with the relevant period. Accordingly it is necessary to determine whether the rule-making procedures subsequent to January 31, 1976 may be given retroactive effect under the factual situation here presented. FEA in its brief sets forth the procedures followed and upon which it relies, which may be summarized as follows:
Among a number of regulatory amendments issued on February 1, 1976 was “a new § 212.85”, the NPCI Last rule. On January 28, 1976, the FEA had issued a notice of public hearing and opportunity for comment on all of its refiner price and allocation rules. On February 27 it issued “another notice of proposed rulemaking and public hearing focusing specifically on § 212.85 and the NPCI Last interpretation for the period prior to February 1, 1976”.79 (DOE Br. p. 47.) A public hearing was held on March 8, 1976, at which most, if not all, of the plaintiffs appeared, and numerous written comments were received. All parties described the oral and written comments as a “storm of protest”.
On April 6, 1976, the FEA issued amendments to its refiner price regulations which, inter alia, repealed § 212.85 retroactively to February 1, 1976, the date on which it was *1061issued. For the period prior to February 1, the FEA stated:
FEA is also still considering appropriate action, including further regulation amendments pursuant to this rulemaking proceeding, regarding the rules on re-coupment of increased non-product costs prior to February 1, 1976. Preamble, 41 Fed.Reg. 15330, 15333 (April 12, 1976).
In a notice published in the Federal Register on August 9, 1976, the FEO [the successor to the FEA] said in part:
Although FEO believes that its interpretation of the meaning of the language used in the regulations is correct — i. e., increased product costs were required to be recovered first; banking of unrecov-ered increased non-product costs was not permitted — it nevertheless acknowledges that refiners might have concluded in good faith that recoupment on a proportional basis was permitted as a result of possibly ambiguous language in § 212.-83(d) and certain information disseminated by the FEA. 41 Fed.Reg. 33282, 33283 (August 9, 1976).
2. Nature of NPCI Last Rule
We agree with both district courts that the NPCI Last recovery rule does not come within the APA exception of an “interpretative rule”. The term “interpretative rule” is not expressly defined in the APA. It is often difficult to determine whether a rule is “substantive” or “interpretative”. In general, a substantive or legislative rule has the force of law; an interpretative rule is merely a clarification of an existing statute or regulation.80 The court must determine into which category the rule falls. The label placed on the rule is not decisive; “rather it is what the Agency does in fact”. Lewis-Mota v. Secretary of Labor, 469 F.2d 478, 481-82 (2 Cir. 1972).
The NPCI Last rule was a “new” rule, not an amendment of an existing rule. It did not merely clarify uncertainty in an existing statute or regulation. The method of sequence of recovery of costs had not been considered in the adoption of any of the pricing regulations. There was no sequence of recovery rule prior to February 1, 1976. As this court noted in Helium, supra, in determining whether a rule is substantive or interpretative the court must look to the basic purpose of the statutory requirements of notice and comment. There can be no question that the agency action here had “palpable effects” upon the regulated industry. Its action should have been exposed “to the test of prior examination and comment”. 569 F.2d at 1145-46.81 Both in the nature of the rule and in its “substantial impact”, the agency action under consideration in these cases is clearly distinguishable from that involved in Energy Reserve Group v. DOE, supra.
3. Compliance with FEAA Rulemaking
We agree with both district courts also that the adoption of the NPCI Last rule without prior notice or opportunity to comment does not satisfy the notice and publication requirements of the FEAA. As the court noted in Standard Oil II, 453 F.Supp. at 243-44, since the rule was “never articulated or published” prior to February 1, 1976, it could not have satisfied the notice and publication requirements of the FEAA unless the “rule falls within an exemption to the rigorous FEAA advanced disclosure requirements”. The only exception under the FEAA, was where “strict compliance is found to cause serious harm or injury to the *1062public health, safety, or welfare, and such finding is set out in detail in such rule [or] regulation ...” FEAA § 7(i)(l)(B), 15 U.S.C. § 766(i)(l)(B). The FEA made no published or unpublished finding in connection with the issuance of its new sequence of recovery rule and did not set out any such finding in its preamble.82 Even if the court could find a compliance with the APA rulemaking requirements, the rule was still invalid for failure to comply with the notice and comment requirements of the FEAA.83
(d) Ban Prohibiting Banking of Non-Product Cost Increases
The ban on banking NPCI was first issued on November 29, 1974, as § 212.-83(e)(4). It was first published on December 5. 39 Fed.Reg. 42368, 42372 (December 5, 1974). There was no attempt to explain the failure to publish the regulation prior to the effective date pursuant to § 7(i)(l)(B) of the FEAA. The FEA contends, however, that under the rules proposed in its notice of September 10, 1974, it was impermissible to bank NPCI, so the issuance of the express ban was “no more than a restatement of the proposed rule’s refusal to authorize NPCI ‘banking’ ”. (DOE Br. at 42.)
The FEA submits that the proposed rules permitted refiners to include in selling prices of covered products the NPCI attributable to sales of those products “in the month of measurement”. § 212.83(c)(l)(ii); 39 Fed.Reg. 32718, 32727 (September 10, 1974). The formulas in that section contained a new factor, M which was defined as the total NPCI in the period “the month of measurement”.84 Since there was nothing in the formulas which provided for inclusion of NPCI from a prior month, the FEA argues that refiners thus had “clear, unambiguous notice” of the ban on banking
NPCI contained in the proposed regulations. The FEA points out that there was no challenge to the validity of the express rule after its promulgation. It argues that if the agency failed to publish the “proposed rule”, that procedural violation was “solely a technical one”. The APA required statement of the rule’s “basis” and “purpose” was contained in the preamble to the November 29, 1974 issuance of the rule; according to the FEA the ban purported “to carry forward the cost absorption features of the previous regulations . . . .” (DOE Br. at 43-45.)
As noted supra, this court has been lenient toward the procedural violations of the embryonic agency, FEO-FEA. See generally our discussion in National Helium, supra, 569 F.2d at 1141—46. In these cases the FEA is asking the court to go one step further. In effect, it claims that the NPCI banking ban was implicit in the proposed rules and then asserts, after this implicit rule was made explicit, that it, in turn, requires the implicit NPCI Last sequence rule.
In these cases, unlike Helium, the questioned rule had a very substantial impact. In many respects it is a crucial link in the FEA’s argument that there was an unstated sequence rule in its overall regulatory structure. Whereas in Helium the questioned rulemaking merely excluded certain materials from the existing regulatory framework, but left the framework essentially unchanged, in the instant cases the rulemaking purported to be a “comprehensive revision” of the Agency’s regulatory framework and, in particular, the Agency clearly intended a radical departure from its prior treatment of NPCI. Thus, the lack of opportunity for “meaningful comment” *1063upon a published, proposed rule truly defeated the educational process that the procedural requisites of the FEAA and APA were intended to promote. See Shell Oil Co., supra, 574 F.2d at 516.
David G. Wilson, Deputy General Counsel of FEA, admitted that the NPCI banking ban was not even considered until “just before the final regulations were issued” to avoid confusion “due to the shift in treatment of non-product costs . . . .” Wilson Depo. 62-63 (R. 1135-36). If the agency did not consider it, how could the refiners have known what was intended? In fact, from the September 10, 1974, notice the refiners may well have had reason to believe that NPCI and PCI were to be treated the same in nearly all respects; the explanation of the new NPCI rules talked about the adoption of “a single method of calculating cost pass through” so that NPCI and PCI pass throughs would be calculated “in the same manner”. 39 Fed.Reg. 32721 (1974) (R. 5174). The only express difference would be that NPCI pass throughs would remain subject to the profit margin limitation.
We recognize that the “month of measurement” language relied on by the FEA does suggest that NPCI cannot be banked, but that is rather weak support for its position when compared to the above language which would have easily led to a contrary view. As the district court held in Phillips Pet. II, at 800, n. 191, section 7(i)(l)(B) of the FEAA required “publication of the rule itself, not ‘clues’ from which a regulated party might deduce the agency’s intent”.85
Since the Agency itself did not consider the need for such a rule until just before it was issued and the refiners made no comment upon the allegedly implicit ban, there could hardly have been any meaningful opportunity for comment prior to the rule’s issuance. The lack of comment on the allegedly implicit ban contrasts sharply with the refiners “nearly unanimous opposition” to the Agency’s simultaneous proposals to limit PCI banking. If through proper notice they had been given an opportunity to comment, they might have put forward the same arguments against the rule later asserted in opposition to the February, 1976, NPCI Last rule — that the NPCI banking ban provided an incentive to reduce refinery operations and instead supply customers with purchased refined products since all cost increases in purchased products could be banked.
The FEA admitted that before the Agency’s treatment of NPCI was revised, refiners could recover their NPCI over a twelve month period. See R. 5787; see also our discussion, supra, at 1049-1050. The pre-notification rules for NPCI pass throughs thus allowed what was tantamount to NPCI banking. Under these circumstances, the departure from prior practice should have been clearly expressed.
Finally, the FEA never considered the basis, purpose, or impact of the NPCI banking ban during either its discussion of the proposed rules or its explanation of the final rules as issued. We agree with both district courts that the FEA violated the procedural requirements of both the APA and FEAA in adopting the NPCI banking ban as well as the NPCI Last rule.
(e) Imposition of NPCI Last Method Retroactively
We are mindful of the holding of the Supreme Court in SEC v. Chenery Corp., 332 U.S. 194, 203, 67 S.Ct. 1575, 1581, 91 L.Ed. 1995 (1947), that agency action “might have a retroactive effect [is] not necessarily fatal to its validity. Every case of first impression has a retroactive effect . .” The validity of the agency action depends upon the “balance” between the “mischief of producing a result which is contrary to a statutory design or to legal and equitable principles” and “the ill effect of the retroactive application of a new standard”. Id. As the court noted in Retail, *1064Wholesale and Department Store Union v. NLRB, 151 U.S.App.D.C. 209, 219, 466 F.2d 380, 390 (1972), “[wjhich side of this balance preponderates is in each case a question of law, resolvable by reviewing courts with no overriding obligation of deference to the agency decision . . . ” The court concluded that among the factors to be considered are “(1) whether the particular case is one of first impression, (2) whether the new rule represents an abrupt departure from well established practice or merely attempts to fill a void in an unsettled area of law, (3) the extent to which the party against whom the new rule is applied relied on the former rule, (4) the degree of the burden which a retroactive order imposes on a party, and (5) the statutory interest in applying a new rule despite the reliance of a party on the old standard”. Id.
Both district courts found no basis for according the February 1,1976 rule retroactive effect. In Phillips Pet. II, the court found three of the five factors set forth above determinative; “(1) the extensive reliance of refiners on the absence of a required NPCI Last method of recovery, (2) the significant burden a retroactive imposition of an NPCI Last rule would have on refiners (the rule would force a reduction of approximately $1.3 billion in refiners’ banks), and (3) the absence of any substantial statutory interest which would be served by applying the rule retroactively given the reliance of the plaintiffs and others on the auditors’ advice to use the proportional method”. 449 F.Supp. at 798.
In Standard Oil II the court first quoted from Judge Jerome Frank’s opinion in To-bin v. Edward S. Wagner Co., Inc., 187 F.2d 977 (2 Cir. 1951), where he said in part:
. [I]t has often been recognized that, as a legislature cannot foresee all possible particular instances to which legislation is to apply, it must therefore be reasonably so interpreted to fill in gaps. But when the legislature delegates to an administrative official the authority, by “sublegislation”, to issue regulations, in order to fill in those gaps, then the regulations, precisely because they particularize, ought not be as generously interpreted as the statute. In fairness to the regulated, the provisions of the regulations should not be deemed to include what the administrator, exercising his delegated power, might have covered but did not cover. 453 F.Supp. at 239.86
The court in Standard Oil II determined that regardless of the FEA’s intent, “its regulations never communicated an all product cost first sequence of increased cost recovery rule to refiners during the January 1, 1975 through January 31, 1976 time frame, and this court will not cure that defect by parroting the strained post hoc ‘interpretation’ which the agency expressed for the first time on February 1,1976”. 453 F.Supp. at 241.
In contending that the district courts erred in denying retroactive application of the NPCI rule, the FEA relies upon a statement of Dean Erwin Griswold that agencies require “freedom in working out the proper construction of a statute in the early days after its enactment”, that changes in the early days may often be made retroactively, and that, “[i]n the absence of long-contin-uedness, the principle of non-retroactivity should yield to the necessities of the process of administrative formulation of a sound rule”. A Summary of the Regulations Problem, 54 Harvard Law Rev. 398, 413 (1941). We do not question Dean Gris-wold’s statement, but we are not persuaded that it is applicable here. The NPCI Last rule, adopted on February 1, 1976, was not adopted in the early days of the agency. Nor can it be said it was necessary in the process of formulating a sound rule. On the contrary, after comments were received, the Agency itself rescinded the rule for future application because of its recognized adverse effects.
In Shell Oil v. FEA, supra, this court, in holding an FEA regulation invalid for fail*1065ure to comply with the requirement of 15 U.S.C. § 766(i)(l)(C) for an opportunity for “an oral presentation of views”, noted that “an opportunity to comment after the rule is promulgated will not suffice”, citing State of Maryland v. Environmental Protection Agency, 530 F.2d 215, 222 (4 Cir. 1975), vacated and remanded, for consideration of possible mootness, 431 U.S. 99, 97 S.Ct. 1635, 52 L.Ed.2d 166 (1977). Shell Oil v. FEA, 574 F.2d at 516, n. 8. In Maryland v. EPA, the court held a rule invalid for failure to comply with notice requirements of the APA and the Clean Air Act, 42 U.S.C. § 1857c-5(c)(l), agreeing with petitioners “that the Agency’s willingness to accept post-promulgation comments after the fact from interested parties is evidence of the confusion attendant to the adoption of certain of the strategies included in the plan and illustrates a lack of prior information”. The court also emphasized “that, in light of the ‘drastic impact’ which compliance” with the regulation would have, “adherence to applicable statutory provisions is necessary”. Maryand v. EPA, 530 F.2d at 222.
That the Agency abandoned its proposed rule prospectively in April, 1976, and admitted in proposing a class exception that refiners may have justifiably relied on erroneous advice87 demonstrates the inequities in imposing the “interpretation” retroactively. Had the NPCI Last rule been in effect during the relevant period, the refiners would admittedly have followed different procedures with respect to reduction of inventories, cutting back domestic refinery output, and avoiding constant price changes. We agree with the district courts that under the undisputed facts the NPCI . Last rule could not properly be given retroactive effect.88
IX. SUMMARY JUDGMENT
We recognize the well settled rule that summary judgment is proper only where there is (1) no genuine issue of any material fact and (2) the prevailing party is entitled to a judgment as a matter of law. Rule 56(c) Fed.R.Civ.Pro. In determining that no issues of material fact exist, the court may not rely solely on the fact that the opposing parties have made cross-motions for summary judgment but must examine the record itself to determine whether material issues of fact exist. Starsky v. Williams, 512 F.2d 109, 112 (9 Cir. 1975).
The party opposing a motion for summary judgment is entitled to have all facts viewed in the light most favorable to it and to all reasonable inferences which may be drawn from the facts. Tyler v. Vickery, 517 F.2d 1089, 1094, reh. denied sub nom., Perry v. Sell, 521 F.2d 815 (5 Cir. 1975), cert. denied 426 U.S. 940, 96 S.Ct. 2660, 49 L.Ed.2d 393 (1976). In determining whether a factual inference which a party seeks to draw is reasonable, the court may look to other evidence in the record which tends to make that inference more or less plausible. Id.
To preclude summary judgment any disputed facts must be material. They must affect the outcome of the litigation. Mobil Oil Corp. v. FEA, 566 F.2d 87, 92 (Em.App.1977). While courts must proceed with caution in determining litigation by summary judgment, “the mere fact that the *1066issues may be complex is not in itself a valid reason to deny summary judgment when there is no material issue of fact and the movant is entitled to judgment as a matter of law”. Zweig v. Hearst Corp., 521 F.2d 1129, 1136 (9 Cir.), cert. denied, 423 U.S. 1025, 96 S.Ct. 469, 46 L.Ed.2d 399 (1975). See also 6 Moore’s Federal Practice H 56.16 at 56-661. “The legal questions do not become easier by postponing resolution until after an unnecessary trial . . Becker v. Safelight Glass Corp., 244 F.Supp. 625, 631 (D.Kan.1965).
Summary judgment is an appropriate mechanism for disposing of legal questions of statutory and regulatory construction. Mobil Oil, 566 F.2d at 92. “Where a decision turns on the meaning of the words in a statute or regulation, a legal question is presented for the court to decide.” International Society for Krishna Consciousness v. Rochford, 425 F.Supp. 734, 738 (N.D.Ill. 1977).
With these principles in mind, we turn to the FEA’s contention that both district courts erred in granting summary judgment in favor of the plaintiff refiners. It is the position of the FEA that “summary judgment would have been an appropriate mechanism had the courts and the plaintiffs limited themselves to looking at the language of the regulations and the procedural issues instead of being influenced overly by all of the refiners’ reliance claims, unsupported as they were”.89
We cannot find that either court was unduly influenced by claimed reliance by the refiners on the interpretation of the regulations made by officials and auditors of the FEA. As noted supra, both courts dismissed allegations in the respective complaints which raised factual issues relating to good faith reliance on the conduct and representations of the FEA officials and to equitable estoppel of the Agency, recognizing that these issues were not ripe for judicial determination.90
The conclusions of the district courts were not based on the refiners’ reliance on the FEA’s interpretation, but rather on a careful analysis of the applicable regulations, which both courts found to be ambiguous and confusing. Interpretations of various FEA representatives and memoranda of their conferences supported that conclusion. Most references in the opinions of the district courts to the refiners’ reliance on the FEA interpretations were to memoran-da of the Agency itself, calling attention to the fact that the refiners had relied upon the interpretations by the Compliance Office and FEA auditors.91
But this reliance was not the basis of the courts’ conclusions that there was no express sequence of recovery rule prior to February 1, 1976; that the regulations did not require the NPCI Last method of recovery; 92 that the FEA did not comply with the required rulemaking procedures; and that the belated interpretation did not qualify for retroactive application.93
*1067In reaching these conclusions the district courts did not rely on any disputed material facts. They properly focused on how the FEA, not the refiners, interpreted the regulations in determining that there was no valid sequence of recovery rule in effect during the relevant period.94 They could properly conclude from undisputed facts that there was uncertainty and confusion on the part of both the FEA and the refiners.95
X. CONSTITUTIONAL ISSUES
The appellees have challenged the constitutionality of the sequence of recovery rule adopted on February 1, 1976, on the grounds that: (1) it would be “[in]consistent with the considerations of fundamental fairness embodied in the due process clause of the Fifth Amendment for the regulations to be retroactively interpreted” to require a NPCI Last sequence of recovery rule; and (2) the NPCI Last sequence of recovery rule “was not set forth in language of sufficient clarity to be understood by men of common intelligence,” — i. e., the rule is void for vagueness.
The Ohio court found the constitutional questions to be substantial and certified them to this court pursuant to 12 U.S.C. § 1904, note § 211(c) (1976). Standard Oil II, 453 F.Supp. at 246. The Delaware court did not reach the constitutional questions in its decision.
Because of our conclusions that the petroleum pricing regulations in effect during the relevant period did not require a particular sequence of recovery and that the *1068FEA’s rule adopted on February 1, 1976 should not be applied retroactively for the relevant period, we conclude it is unnecessary to consider the constitutional issues raised by the appellees.
XI. UNION OIL COMPANY DOCUMENT
On July 28, 1978 the FEA moved this court for leave to file a supplemental brief to admit into the record a document which it had recently obtained through discovery in unrelated litigation in the United States District Court for the District of Minnesota.96 The document, a confidential, internal memorandum, dated May 29, 1975, between officials of Union Oil Company of California (Union Oil) discussed Union Oil’s perception of the cost pass through and recovery provisions of the petroleum pricing regulations and recommended a strategy of recovering all its non-product costs first.97
The FEA contends that the reasoning in the document is “representative of how one major refiner in fact approached the regulations, and probably how the industry in general approached the regulations . . . .” As well, it is “persuasive evidence” (1) that the plaintiff-refiners’ “self-serving allegations about their contemporaneous misunderstanding” of the FEA regulations should be entitled to little or no weight; (2) that “refiners were not denied due process or proper notice of the NPCI Last rule”, (3) that “it was never ‘settled law’ that refiners were not required to recover their NPCI Last”, and (4) that refiners “may have used” the NPCI First or proportional methods of cost recovery “to circumvent” the NPCI banking ban and the profit margin limitation rule.98
For the following reasons we deny the FEA’s motion to admit the Union Oil memorandum. First, Union Oil is not a party to this appeal, nor has it ever been involved in these law suits. While the memorandum may represent Union Oil’s strategy concerning its approach to the petroleum pricing regulations, that strategy cannot be imputed to other refiners. Second, the memorandum was never before either of the district courts. It is well settled that facts not presented in the district court may not be asserted on appeal. Coalition for Responsible Regional Development v. Brinegar, 518 F.2d 522, 527 (4th Cir. 1975); Clark Marine Corp. v. Cargill, Inc., 345 F.2d 79 (5th Cir. 1965), cert. denied, 382 U.S. 1011, 86 S.Ct. 620, 15 L.Ed.2d 526 (1966). If the FEA had wished to introduce facts relevant to the consequences of a NPCI Last rule which would demonstrate a genuine issue of material fact, it was bound to do so at the district court level. See Garcia v. American Marine Corp., 432 F.2d 6 (5th Cir. 1970). Third, we question the relevance of this memorandum to the legal interpretation of the regulations. As the FEA stated in its brief on this motion: “The only proper way to construe the regulations is to look first at their language and then at the preambles and other relevant official public documents surrounding their promulgation.”99 Judge Larson of the Minnesota District Court also recognized this when he ruled on the FEA’s motion to obtain the document:
*1069THE COURT: As a practical matter, I am not certain what value this would have to the Temporary Emergency Court of Appeals. You might have 150 varying interpretations by oil companies, for example, and what value would that be? The regulations are there. The problem is “What do the regulations say?” 100
The FEA’s motion to admit into the record the memorandum of the Union Oil Company of California is denied.
XII. CONCLUSION
We conclude that: (1) the questions decided by the district court were ripe for judicial resolution; (2) the regulations in effect during the relevant period — -January 1, 1975 to January 31, 1976 — were ambiguous and confusing with respect to sequence of recovery of product and non-product costs; (3) the FEA did not issue any explicit sequence of recovery rule until February 1, 1976; (4) the NPCI Last method was a reasonable interpretation of existing regulations and was within the statutory authority of the FEA; but (5) this interpretation of existing regulations was not compelled nor required; (6) the FEA did not contemporaneously construe the regulations during the relevant period as requiring a NPCI Last sequence of recovery rule, and there was in fact no required sequence of recovery rule during the relevant period; (7) there was no basis for according deference to the FEA interpretation first announced on February 1, 1976; (8) the FEA failed to comply with rule-making requirements of both the Administrative Procedure Act and the Federal Energy Administration Act with respect to both (a) the NPCI Last sequence of recovery rule and (b) the ban prohibiting the banking of non-product cost increases; (9) the sequence of recovery rule first adopted on February 1, 1976 was not entitled to retroactive application for the period January 1, 1975 to January 81, 1976; and (10) the district courts did not resolve any disputed material facts and the appel-lees were entitled to a judgment as a matter of law. Accordingly we affirm the summary judgments entered by both district courts,
. See 41 Fed.Reg. 5111, 5113 (February 4, 1976).
. FEA became a part of the Department of Energy on October 1, 1977, 42 U.S.C.A. § 7151(a); Executive Order No. 12009, 42 Fed. Reg. 46267 (September 15, 1977). Since these cases involved actions of the former agency, this opinion will refer generally to the FEA, as did both district courts in their opinions.
. See Standard Oil Co. v. Federal Energy Administration (Standard Oil I), 440 F.Supp. 328, 365, 372 (N.D. Ohio 1977); Phillips Petroleum Co. v. Federal Energy Administration (Phillips Pet. I), 435 F.Supp. 1239, 1249 (D.Del.1977).
*1034The district courts also granted motions to dismiss the complaints as to allegations of good faith reliance on the representations and conduct of FEA officials, equitable estoppel of the agency, and lack of a “rational basis” for applicable regulations.
. Regulations permitted refiners to charge a “base price” equal to the sum of the weighted average price on May 15, 1973, and the increased product cost incurred between May 1973, and the month of measurement. An amendment to the pricing regulations, adopted by the Cost of Living Council on September 12, 1973, provided that product cost increases not recouped in their entirety in one month could be carried over or “banked” for a later month. There was no such provision for banking non-product costs, and in December, 1974 a new provision expressly prohibited the banking of non-product cost increases. These regulations and regulations defining “base price” and authorizing pass through increases for non-product costs will be discussed later in this opinion.
. During the relevant period, 29 refiners, including Exxon, Gulf, Texaco, and Sohio, plaintiffs in the Ohio cases, used the NPCI first method. Approximately 29 refiners, including Shell, Amoco, and the six Delaware plaintiffs used the proportional method. Mobil, Sun, and Arco at different times utilized both methods. Approximately 46 small, independent refiners, not parties to these actions, did not pass through any non-product cost increases or recovered these increases after all of their product cost increases had been recovered. See Standard Oil I, 440 F.Supp. at 344; 41 Fed.Reg. 40559, 40560 (September 20, 1976).
. The provisions of the applicable statutes and regulations are set forth in more detail in the district court’s opinions. See Standard Oil I, 440 F.Supp. at 332-337; Standard Oil II, 453 F.Supp. at 206-211; Phillips Pet. II, 449 F.Supp. at 765-769.
. Prior to the final adoption of the proposed special rules, refiners were subject to “Special Rule No. 1”, which the CLC had issued on March 6, 1973. That rule introduced mandatory price controls for refiners that had annual sales of $250 million or more. It also imposed a profit margin limitation and established a pre-notification procedure for companies that sought to increase prices more than one and one-half percent above their base prices. Standard Oil II, 453 F.Supp. at 206; 38 Fed.Reg. 6283 (March 8, 1973).
. “Base price” was defined in 6 C.F.R. § 150.-358(g) as follows:
The base price for sales of an item by a refiner is the weighted average price at which that item was lawfully priced in transactions on May 15, 1973, or, if none occurred on that date, in the transaction next preceding May 15, 1973, plus increased costs of imports incurred after May 15, 1973, and measured pursuant to § 150.356(b) and increased costs of domestic crude petroleum incurred after May 15, 1973, and measured pursuant to § 150.357(b).
38 Fed.Reg. 22536, 22541 (August 22, 1973).
. § 150.356(c)(1) was amended to read in pertinent part:
“If in any month ... a firm establishes a base price for any covered product which does not include the entire amount of increased costs calculated pursuant to this formula . . . the unused portion may be added to the May 15, 1973, selling price to compute the respective base price . . . for a subsequent month.” 38 Fed.Reg. at 25688. The amendment also provided express mathematical formulas for calculating base prices and maximum selling prices.
. Executive Order No. 11748, 38 Fed.Reg. 33575 (December 4, 1973).
. 39 Fed.Reg. 744, 761 (January 2, 1974). These regulations were later renumbered and republished in Title 10 of the Code of Federal Regulations. 39 Fed.Reg. 1924 (January 15, 1974).
. Executive Order No. 11790, 39 Fed.Reg. 23185 (June 25, 1974).
. Phillips Pet. I contains the following illustration of the use of the respective methods:
For example, assume that in a given month a refiner had incurred increased product costs of $70 and increased non-product costs of $30, a total of $100. Assume, however, that the refiner actually recovered during the following month only $90 (that is, 90 percent) of its increased costs. (This could happen, for example, because the refiner misestimat-ed its sales when setting prices or because the market precluded charging the maximum allowable sales price.) Under the NPCI First Method, the refiner would be deemed to have recovered $30 in non-product costs and $60 in product costs; it could bank the remaining $10 of product costs for recovery in later
months. Under the NPCI Last Method, the refiner would be deemed to have recovered $70 in product costs and $20 in non-product costs; it would lose forever the remaining $10 of non-product costs. Under the Proportional Method, the refiner would be deemed to have recovered $63 in product costs ($70 times 90 percent) and $27 ($30 times 90 percent) in non-product costs; it could bank $7 of product costs for recovery in later months and would forever lose the remaining $3 of non-product costs.
435 F.Supp. at 1243, n.5.
. The directives and communications are described in detail in Standard Oil II, 453 F.Supp. at 211-17, and infra at 1052-1055.
. The communications between Congressman Dingell and his staff and various FEA officials are described in Standard Oil I, 440 F.Supp. at 344-47, and Standard Oil II, 453 F.Supp. at 220-21.
. 41 Fed.Reg. 43953 (October 5, 1976).
. The FEA argues that the Ohio court “appears to have sided with the agency” on the cost recovery sequence issue and contends that “had the Ohio court concluded that all PCI had to be included in base prices each month, then it would have been compelled to sustain the agency’s interpretation”. (DOE Br. at 19-20). This is an inaccurate characterization of the Ohio court’s decision. That court found that nothing in the definition of base price “was crafted to embrace an unarticulated sequency of recovery rule”. Standard Oil II, 453 F.Supp. at 207. It concluded that the regulatory history was not persuasive to establish the FEA’s “assertion that the maximum price rule also specified a sequence of cost increase recovery". Id. at 222. (Emphasis in original).
. “Covered products” were products described in the 1972 edition, Standard Industrial Classification Manual, Industry Code 1311 (except natural gas), 1321 or 2911. § 150.352; 38 Fed. Reg. 22536, 22538 (August 22, 1973). The CLC later put gasoline, No. 2 diesel fuel, and No. 2 heating oil into a subcategory of covered products called “special products”, which received different treatment than the remaining covered products. § 150.356(b), 39 Fed.Reg. 25686, 25688 (September 14, 1973).
. See Standard Oil II, 453 F.Supp. at 223.
. Id. Those costs other than costs related to imported crude oil and petroleum products, and domestic crude oil were later labelled as “non-product” costs. § 212.83(b), 39 Fed.Reg. 42368, 42369 (December 5, 1974).
. “Cost of Petroleum Product” was defined as “(1) For purposes of domestic petroleum products other than crude petroleum, the purchase price including transportation costs. (2) For purposes of imported petroleum products other than crude petroleum, the landed cost.” § 212.83(b); 39 Fed.Reg. 42368, 42369 (December 5, 1974).
. A second formula computed the total amount of product cost increases that could be added to the May 15, 1973 selling prices of covered products other than special products to determine the base prices of those products. This amount was represented by the figure “Dju”. § 212.83(c)(2)(ii); 39 Fed.Reg. 42368, 42370 (December 5,1974).
. A revised § 150.356(c) was adopted by the CLC in September, 1973. 38 Fed.Reg. 25686, 25688 (September 14, 1973). It provided in pertinent part:
(c) Allocation of increased costs — (1) General Rule. In computing its ceiling prices and base prices for covered products pursuant to §§ 150.355(c) and 150.358(g), a refiner may increase its May 15, 1973, selling prices to each class of purchaser on the first day of each month beginning with September 1973 by an amount to reflect the increased costs of imports and increased costs of domestic crude petroleum allowable under paragraphs (d), (e), and (f) of this section, provided that the amount of increased costs used in computing a ceiling price or base price is calculated by use of the general formula set forth in paragraph (c)(2) of this section.
This subsection was the forerunner of § 212.-83(c)(l)(i), the allocation of increased costs section in effect during the relevant period. Although not identical to § 212.83(c)(l)(i), § 150.-356(c)(1) was very similar to it. Section 212.-83(c)(1) provided in pertinent part:
(c) Allocation of increased product costs— (1) General rule — (i) Special products. In computing base prices for sales of a special product, a refiner may increase its May 15, 1973 selling prices to each class of purchaser once each calendar month beginning with November 1973 by an amount to reflect the increased product costs attributable to sales of that special product using the differential between the month of measurement and the month of May, 1973 provided that the amount of increased costs used in computing a base price is calculated by use of the general formula set forth in paragraph (c)(2)(i) of this section.
39 Fed.Reg. 42368, 42370 (December 5, 1974).
. 38 Fed.Reg. 25686, 25688 (September 14, 1973). This provision was amended in November, 1973, “to address situations where a refiner’s estimated figures under the formula produce actual revenue in excess of that permissible”. Preamble, 38 Fed.Reg. 30267, 30268 (November 2, 1973). The banking provision was put into a new subsection, which provided in part:
(d) Carryover of costs, (i) If in any month ., a firm charges prices for a special product . . which result in the recoupment of less total revenues than the entire amount of increased costs of imports and increased costs of domestic crude petroleum calculated for that product . . . and that unused amount of increased costs is not used to increase May 15, 1973, selling prices pursuant to paragraph (c)(l)(ii) of this section [that is, allocated to covered products other than special products], the amount of increased costs not recouped may be added to the May 15, 1973, selling prices to compute the base prices for that special product for a subsequent month. If in any month . ., a firm charges prices for a special product . . . which result in the recoupment of more total revenues than the entire amount of increased costs of imports and increased costs of domestic crude petroleum calculated for that product . . ., the amount of excess costs recouped must be subtracted from the May 15, 1973, selling prices to compute the base prices for that special product ... for the subsequent month.
38 Fed.Reg. at 30271. (Emphasis added.)
. The refiners point to additional language in the preamble to technical amendments of the banking provision which they allege demonstrates the FEA intended the base price rule to be flexible. In that preamble the FEA identified the formulas in § 212.83 as the formulas “by which a refiner must calculate certain product costs which may be used to determine its base prices for covered products”. Preamble, 39 Fed.Reg. 15138 (May 1, 1974) (Emphasis added.) In its brief the FEA admits that there is “ambiguity inherent in the use of the word ‘may’ ”, but argues that it is “reasonable” to read this language as meaning that only the amount of PCI calculated by use of the formulas “may” be used to compute base prices. (DOE Reply Br. at 14). We think, however, that the refiners’ interpretation of this language is also reasonable and note that while the FEA argues that its interpretation of this preamble is reasonable, it does not argue that it is compelled.
. The FEA argues that there is no provision in the regulations, the formulas, or Form FEO-96 (a “Monthly Cost Allocation Report” refiners were required to complete to report their product cost increase allocation) that permitted holding back costs or allocating costs directly to banks. As discussed supra, whether the regulations and formulas permitted discretionary allocation of PCI to May 15, 1973, selling prices depended upon one’s interpretation of the word “may”. If viewed as discretionary, there is nothing in Form FEO-96 which prohibited a refiner from including less than all of his available PCI in the base price computation. The final calculation requested by Form FEO-96 was entitled “Maximum permissible price adjustment to each May 15, 1973 selling price of gasoline”. Form FEO-96, Part IV, (26); (R. 2738). We think the “maximum permissible price adjustment” could reasonably have been interpreted as just that: the maximum amount of increased product cost attributable to gasoline that could be added to the May 15, 1973 selling price. Nothing in Form FEO-96 stated that the maximum amount was also the amount that must be added to the May 15, 1973 price.
. The banking provision as initially adopted in September, 1973, was later amended in November, 1973. See n. 24, supra, for both versions.
. In November, 1973, the banking provision was amended and put into a new subsection. See n. 24, supra. The FEA argues that the Ohio court ignored this amendment which “made clear that banking was permitted only when a refiner recouped less increased revenues than the total amount of PCI allocated to base prices at the beginning of the month pursuant to the PCI allocation formula”. (DOE Br. at 25-26) (Emphasis added.) The preamble to a proposed modification of the banking provision in September, 1974, however, belies this interpretation of the banking provision as it existed from November, 1973 through December, 1974. In the preamble the FEA stated that the banking provision was intended to serve two primary functions, one of which was to permit refiners to smooth out price adjustments over a period of time by voluntarily delaying inclusion of increased product costs in their base prices. Preamble, 39 Fed.Reg. 32718, 32723 (September 10, 1974). This is exactly the kind of discretion the refiners argued that the base price rule allowed them to exercise.
. At that time refiners were required to pre-notify the FEA of any proposed price increases in excess of base price. § 212.121; 39 Fed.Reg. 1924, 1960 (January 15, 1974). The FEA could disapprove, modify, suspend, or defer the proposed price increase, § 212.124; 39 Fed.Reg. at 1960-61. In December, 1974, the FEA abolished the pre-notification requirements effective January 1, 1975. 39 Fed.Reg. 42368 (December 5, 1974).
. The condition stated:
No part of the increase above base prices of covered products authorized by this Order may be applied to the price of a covered product below the base price. In applying any permitted increase above a base price, no part of the increment so applied may be used to support a calculation permitted or required by § 212.83.
Affidavit of Thomas F. Músico, fl 7 (September 14, 1977) (R. 2537, 2541).
. For our discussion of this decision as it relates to the sequence of recovery of increased costs, see infra, at pp. 1045-1046.
. During 1974 the FEA allowed Standard Oil of Ohio (Sohio) to use non-product cost increases to make upward price adjustments under the pre-notification procedure when Sohio had not passed through all of its product cost increases. See Affidavit of Donald O. Maxwell, H 7 — 11 (September 14, 1977) (R. 2425, 2426-28). Exhibit C of that affidavit is an FEA Decision and Order granting such a pre-notified non-product cost passthrough.
In his deposition, Gorman C. Smith, the FEA’s Assistant Administrator for regulatory programs, testified that although his understanding of the definition of base price would have required a refiner to exhaust its product costs bank prior to charging a price in excess of base price, he did not know if anyone in the FEA ever checked during the pre-notification period to determine whether a refiner’s product cost bank had been exhausted before permitting recovery of non-product cost increases. Smith Depo. 513 (R. 2079). More important, David G. Wilson, a Deputy General Counsel with the FEA responsible for drafting and interpreting many of the regulations in question, knew of no provision in the regulations requiring exhaustion of product cost banks as a condition to implementing price increases under the pre-notification procedure. Wilson Depo. 121 (R. 1195). This further demonstrates the confusion which reigned in the FEA over the interpretation of its regulations.
. As support for its conclusion that the base price was not fixed, the Ohio court, as did the Ohio plaintiffs, relied in part upon an official Ruling of the FEA, Ruling 1974-12 (May 28, 1974) (R. 3122). That ruling involved an example of how refiners could allocate increased costs attributable to the production of jet fuel. Jet fuel, however, belonged to the category of “covered products other than special products”, while gasoline was a “special product”. The allocation formulas for covered products other than special products differed from the allocation formulas for special products. Compare § 212.83(c)(l)(i) to § 212.83(c)(l)(ii); 39 Fed.Reg. 1924, 1955 (January 15, 1974). Therefore, we do not rely upon Ruling 1974-12.
. Section 212.82(c) provided in part:
(c) Allowable price in excess of the base price.
A refiner may only charge a price in excess of the base price of a covered product in order to recover on a dollar for dollar basis increased nonproduct costs incurred between the month of measurement and the month of May 1973 and measured pursuant to the provisions of § 212.81.
39 Fed.Reg. 42368, 42369 (December 5, 1974). Although the wording of this subsection changed with amendments, the intent remained the same from the first adoption of the petroleum pricing regulations in 1973 through the end of the relevant period in 1976. Compare § 150.358(c)(1), 38 Fed.Reg. 22536, 22541 (August 22, 1973) with § 212.82(b)(1), 39 Fed.Reg. 1924, 1952 (January 15, 1974).
. The profit margin limitation was part of the petroleum pricing regulations as initially adopted in August, 1973. § 150.358(j); 38 Fed.Reg. 22536, 22542 (August 22, 1973). It remained part of the regulations until the FEA repealed it in March, 1976. 41 Fed.Reg. 9087 (March 3, 1976).
. The regulation in question was § 212.83(d) entitled “Allocation of increased non-product costs.” It was adopted December 1, 1974, and provided in pertinent part:
*1047In computing allowable prices in excess of base prices for sales of a covered product, a refiner may increase its price above base price . . . each month . . . by an amount to reflect the increased non-product costs attributable to sales of covered products . . . provided that the total amount of increased non-product costs which can be used in computing allowable prices above base prices is calculated pursuant to § 212.87, and provided that the amount of increased non-product costs included in computing allowable prices above base prices of a particular covered product bears the same ratio to the amount of increased product costs included in the base price of that product as the ratio of the total amount of increased non-product cost computed under § 212.87 for the month of measurement bears to the total amount of increased product cost available for allocation to base prices in the current month, pursuant to paragraphs (c) and (e) of this section.
. Section 212.83(e)(1) provided:
(e) Carryover of costs. (1) If in any month . a firm charges prices for a special product which result in the recoupment of less total revenues than the entire amount of increased product costs calculated for that product pursuant to the general formula and allowable under paragraph (c)(l)(i) of this section and that unused amount of increased costs is not used to increase May 15, 1973
selling prices pursuant to paragraph (c)(l)(ii) of this section, the amount of increased product cost not recouped may be added to the May 15, 1973 selling prices to compute the base prices for that special product for a subsequent month. . . If in any month a firm charges prices for a special product which result in the recoupment of more total revenues than the entire amount of increased product costs calculated for that product pursuant to the general formula and allowable under paragraph (c)(l)(i) of this section the amount of excess product costs recouped must be subtracted from the May 15, 1973 selling prices to compute the base prices for that special product for the subsequent month.
39 Fed.Reg. 42368, 42372 (December 5, 1974).
When the product cost banking provision was first adopted it was in § 150.356(c)(1); 38 Fed.Reg. 25686, 25688 (September 10, 1973). In November, 1973, it was amended and incorporated in § 150.356(d); 38 Fed.Reg. 30267, 30271 (November 2, 1973). See supra, n. 24. The provision was again amended in December, 1973 (38 Fed.Reg. 33577, 33581 (December 6, 1973)). When the FEO renumbered the regulations, § 150.356(d) became § 212.83(d). 39 Fed.Reg. 1924, 1956 (January 15, 1974). In December, 1974 § 212.83(d) became § 212.-83(e). 39 Fed.Reg. 42368, 42372 (December 5, 1974).
. See R. 2736.
. See R. 2612.
. 39 Fed.Reg. 32718, 32721 (September 10, 1974).
. 39 Fed.Reg. 42368 (December 5, 1974).
. During 1975 FEA officials met numerous times to discuss revisions of Form FEO-96. At the same time FEA officials distributed worksheets to its field auditors to supplement Form FEO-96 until a new form could be completed. These worksheets directed refiners to use a proportional method of cost recovery rather than the NPCI Last method which the FEA now argues was compelled by the regulations. See our discussion infra, at 1052-1054.
. See supra, n. 37.
. See supra, n. 37.
. See infra, at 1061-1063.
. Wilson Depo. 25-26, 40 (R. 1073, 1098-99, 1113).
. The rules for computing allowable non-product cost increases for special products and for covered products other than special products were in §§ 212.82(c)(2)(i) and (ii), 39 Fed.Reg. 1924, 1953 (January 15, 1974). The rules for computing allowable product cost increases for these categories were in §§ 212.83(c)(l)(i) and (ii), 39 Fed.Reg. at 1955. For a more detailed discussion of the different treatments accorded non-product and product cost increases, see Phillips Pet. II, 449 F.Supp. at 777-78. See also Preamble, 39 Fed.Reg. 32718, 32721 (September 10, 1974).
. Compare the definitions of “diu” (special products) and “Diu” (covered products other than special products) in §§ 212.83(c)(2)(i) and (ii), 39 Fed.Reg. 1924, 1955 (January 15, 1974), with the definitions of “die” (special products) and “Die” (covered products other than special products) in §§ 212.82(c)(3)(i) and (ii), 39 Fed. Reg. at 1953.
As the Ohio court noted, § 212.82(c)(2)(i), in effect prior to December, 1974, “permitted a refiner to calculate a percentage increase in non-product costs, pre-notify this ‘rate’ of increase, and then use the rate in recovering increased non-product costs in the forthcoming 12-month period”. 453 F.Supp. at 234.
. See § 212.82(d); 39 Fed.Reg. 1924, 1953 (January 15, 1974).
. On December 1, 1974, the FEA deleted the requirement that refiners pre-notify their proposed non-product cost increases and amended the pricing formula for calculating the allowable price increase for non-product costs. Section 212.83(d), as amended, provided that increased non-product costs which could be used in computing allowable prices above base prices were to be calculated according to § 212.87, provided that the amount of increased non-product costs included in computing the allowable prices above base prices bore a particular ratio to the amount of increased product costs included in the base price.
The formula in § 212.87 for computing non-product cost increases defined allowable non-product costs increases used to justify a price in excess of base price as a certain proportion of a refiners’ non-product cost increases. § 212.87(b)(1); 39 Fed.Reg. 42368, 42373 (December 5, 1974). Refiners were required to compute their increased non-product costs for each defined category of non-product cost (e. g., refinery fuel cost increase, labor cost increase). Only the net increase in non-product costs for a particular month could be used to justify an allowable price in excess of base price. § 212.87(b)(2); 39 Fed.Reg. at 43273.
Many refiners and FEA officials interpreted §§ 212.87(b) and 212.83(d) as requiring a proportional method of recovery rather than a NPCI Last method of recovery. See text accompanying n. 36; see also, infra, at 1052-1055.
For the text of § 212.83(d), see supra, n. 36.
. 41 Fed.Reg. 15330, 15331 (April 12, 1976).
. The FEA contends this “erroneous” advice from its lower level officials is relevant only to the question of refiners’ detrimental reliance, a matter it is taking up in individual exception proceedings.
. See Standard Oil II, 453 F.Supp. at 211-217, and Phillips Pet. II, 449 F.Supp. at 785-92.
. See R. 2752.
. See supra, n. 36.
. Affidavit of Andrew H. Drance, Jr., ¶s 6-7 (R. 3476, 3479-81).
. See R. 2752, 2754. Before he drafted the Change Notice, Drance contacted Peter Luedtke, then the FEA’s Assistant General Counsel for Pricing, to check out the proportional cost recovery theory. Drance and Luedtke have different recollections about the conversation. In a very brief conversation, Drance advised Luedtke that he was instructing the field auditors to use the proportional method of recovery. After Drance explained his understanding of § 212.83(d), he thought, Luedtke responded, “that’s what it should have said”. Drance aff’d, ¶ 8, R. 3481. Luedtke testified he thought Drance was talking about the allocation of costs among product categories rather than a method of cost recovery. Luedtke Depo. 46-47; R. 7470, 7588. In light of all the other evidence of contemporaneous construction, we view this disputed conversation as inconsequential.
. A memorandum prepared by Fred Laughlin, a consultant from the Price, Waterhouse accounting firm, who had been hired to advise the FEA in its revision of Form FEO-96, summarized the discussion of cost recoveries at this meeting:
During the discussion of special issues, it was noted that the regulations, although specifying that non-product cost increases cannot be carried over, do not specify in what sequence cost increases are recovered. The result is refiners many [sic] consider non-product costs to be recovered first in any month and product costs second because these increases can be carried over, if not fully recovered. It was noted that the FEA is currently telling refiners that these costs are recovered pro rata in the same proportion as the cost increases.
The FEA argues that this memorandum cannot be used as evidence of its contemporaneous construction because it was not the statement of an FEA official and was not made available to the FEA until this litigation had commenced. Additionally, the FEA has submitted affidavits from Mayo Lee and Andrew Drance refuting portions of the memorandum. They state they do not recall anyone at the meeting arguing that non-product costs could be recovered first. Affidavit of W. Mayo Lee, fl 16 (November 18, 1977) (R. 5824, 5833-34); Supplemental Affidavit of Andrew H. Drance, Jr., H 3 (November 17, 1977) (R. 5969).
Because this memorandum was prepared by a person present at the meeting who was hired by the FEA to aid it in redrafting Form FEO-96, it is relevant evidence of what occurred at that meeting. Regardless of whether anyone thought the regulations would allow NPCI to be recovered first, it is undisputed FEA officials present advocated both the NPCI Last and proportional methods.
. 40 Fed.Reg. at 19132.
. Affidavit of Andrew Drance, Jr., fl 15 (October 27, 1977) (R. 3476, 3486).
. Memorandum p. 2 (April 11, 1975) (R. 3526-27).
. 40 Fed.Reg. 40834 (September 4, 1975) (R. 2846).
. See R. 3592-96. Harvey also said in part:
Compliance guidance to the field also understood the interpretation of pro-rata recoveries below base price and issued guidance to this effect by providing a worksheet to compute non-product cost recoveries.
FEA may be committed to that interpretation for the period that the guidance has been available, but we could issue a ruling which changes that interpretation on a prospective basis. It would seem unfair to require refiners to make a retroactive adjustment when they followed our guidance in good faith.
Id. 3594-95.
. See Shell Oil Company Letter of February 24, 1975 (R. 3924, 3925).
. See R. 3921. When Mobil did not hear from the FEA by July, 1975, it resubmitted its request for an interpretation of these regulations. Neither of Mobil’s requests was answered prior to the time the FEA adopted the NPCI Last sequence of cost recovery on February 1, 1976. See Standard Oil II, 453 F.Supp. at 215-16.
. See Amoco Oil Company Letter of August 13, 1975 (R. 3929).
. 41 Fed.Reg. 1680 (January 9, 1976).
. See § 212.85; 41 Fed.Reg. 5111, 5120 (February 4, 1976).
. See Preamble, 41 Fed.Reg. 5111, 5113 (February 4, 1976).
. 41 Fed.Reg. 15330, 15331 (April 12, 1976).
. In California v. Simon, the state and two municipalities challenged the FEO’s rescission of state and local government exemption of crude oil from price controls. In explanation of its ruling, the court said: “We conclude that the procedure in question from the standpoint of substantial statutory compliance, fairness, circumspection, prior notice, opportunity to be heard and abundant record justification, against the background of an embryonic agency faced with complicated takeover tasks under a Congressional deadline, was such as to properly render such mere technical objection unavailing.” 504 F.2d at 440.
. See also Consumer's Union v. Sawhill, 393 F.Supp. 639 (D.D.C.1975), where the district court held invalid a price regulation for unleaded gasoline promulgated without public notice and opportunity to comment. This court affirmed. Consumers Union v. Zarb, 523 F.2d 1404 (Em.App. 1975).
. In National Helium, the court concluded that failure to give notice of amendments to price regulations for natural gasoline prior to publication of the amendments did not render the amendments invalid because the normal rule-making procedure was impracticable, the amendments did not alter the existing regulatory framework, and there was no detrimental impact on the rights of the parties.
. We recognized, however, that “When the questioned agency action does not jeopardize substantive rights or have a ‘substantial impact’ on the purportedly regulated parties, or ‘create substantive effects upon the rights of the parties not involved in the proceedings,’ then lack of compliance with the prior notice and comment requirements of the APA is not fatal. Pennsylvania v. United States, 361 F.Supp. 208, 220-22 (N.D.Pa.) (three judge court), aff'd, 414 U.S. 1017, 94 S.Ct. 440, 38 L.Ed.2d 310 (1973).” 569 F.2d at 1146.
. FEAA 7(i)(l)(B) provided further that the publication, notice and comment requirements “may be waived where strict compliance is found to cause serious harm or injury to the public health, safety, or welfare, and such finding is set out in detail in such rule [or] regulation . . . .” It will be noted that this waiver provision differs substantially from the exception provisions of the APA.
For a detailed comparison of the notice requirements of the APA and FEAA, see Shell Oil Company v. FEA, 440 F.Supp. 876, 881-82, n. 25 (D.Del.1977), aff'd, 574 F.2d 512 (Em.App. 1978). The district court concluded that the FEA had not complied with the FEAA regulations and accordingly it was unnecessary “to decide at this time whatever additional impact the APA exerts on the activities of the FEA”. Id. at 884. This question was not discussed by this court on appeal. As set forth infra, we concluded that there was a failure to comply with the FEAA notice requirements.
. We referred also to remarks of Representative Broyhill in explaining to the House an amendment which became § 7(i)(l). Cong. Rec.H. 1514-15 (March 6, 1974). Representative Broyhill said in part:
Legislation which in the past has attempted to meet emergency or crisis situations has failed to provide for suitable Administrative Procedural provisions. The goal of procedures in this type of legislation should be to provide orderly process for affected individuals and corporations while at the same time providing the greatest degree of safeguards to those individuals and corporations affected by the actions of the agency consistent with the necessity for prompt agency actions.
In the past agencies such as the Cost of Living Council have acted without the benefit [of the views] of those who will be most affected by an action. Often the prior views of those knowledgeable in the field being regulated could have prevented an ill-conceived plan or rule from going into effect — a plan or rule which often had to be subsequently altered. The functions of the agency can best be [carried] out by an Administrator who issues rules, regulations, or orders which emanate from the greatest possible data base. 120 Cong.Rec. 5459 (1974).
. Judge Christensen in his concurring opinion rejected the conclusion that the “substantial impact” test may never be employed in determining whether a rule is interpretative or legislative. He said in part: “Whether a regulation is a mere ‘interpretation’ under some circumstances cannot be evaluated confidently without consideration of its effect upon the industry and the public from the viewpoints of essential fairness, due process, departure from established practices and financial or other consequences.”
Judge Zirpoli dissented by reason, inter alia, of the “substantial impact” he found the regulation had on the parties and his conclusion that the ruling was legislative rather than interpretative and subject to formal rulemaking procedures under both the APA and FEAA.
. The publication of the “proposed interpretation” did not refer specially to retroactive application of its new rule. Rather the FEA relied upon the statement in the preamble that the order specified in the new rule was the same as that under previous regulations. Deputy General Counsel Wilson testified in his deposition that he believed the possibility of a retroactive rulemaking prior to February 1, 1976, was “discussed in connection with the final decision to issue the proposed class exception”, but it was decided that “procedurally it would be better to go the proposed class exception route than to issue a proposed retroactive rulemaking”. Wilson Depo. 105 (R. 1178).
. 41 Fed.Reg. 9199 (March 3, 1976). As noted supra, the FEA conceded in the preamble that the FEA’s previously “proposed regulations were silent on the sequence of recoupment”, but concluded that the regulations implicitly embodied an identical “all products costs first” sequence rule.
. See, e. g., Gibson Wine Co. v. Snyder, 90 U.S.App.D.C. 135, 194 F.2d 329 (1952); Lewis-Mota v. Secretary of Labor, supra, 469 F.2d at 481-82; Pickus v. United States Board of Parole, 165 U.S.App.D.C. 284, 507 F.2d 1107 (1974).
. As noted by the district court in Phillips Pet. II, 449 F.Supp. at 799, n. 184, in St. Francis Memorial Hospital v. Weinberger, 413 F.Supp. 323, 328-9 (N.D.Cal.1976), the court considered the existence of a “genuine ground for a difference of opinion on the wisdom of the policy embodied in the rule” to be another factor which precludes classification of a rule as interpretative. We agree with the district court that the “debate within the FEA throughout the relevant period and the storm of protest” when the rule was published demonstrate the existence of genuine policy issues.
. See 41 Fed.Reg. 5111 et seq. (February 4, 1976).
. Moreover, this is not a case where the district courts have imposed procedural rights in addition to the statutory requirements. It is clear that this may not be done. The discretion of administrative agencies and not that of the courts is to be exercised in determining when procedural devices in addition to those specified in the applicable statutes should be employed. Vermont Yankee Nuclear Power Corp. v. NRDC, 435 U.S. 519, 98 S.Ct. 1197, 55 L.Ed.2d 460 (1978). We are concerned here with compliance with the minimum statutory requirements.
. § 212.83(c)(3); 39 Fed.Reg. 32718, 32728 (September 10, 1974). The M factor was not a part of the formula finally adopted.
. See also Rodway v. USDA, 168 U.S.App. D.C. 387, 392, 514 F.2d 809, 814 (1975), where the court held that a broad notice of intention to revise regulations governing the Food Stamp Program was insufficient to include the allotment system “by inference”.
. See also Judge Manos’ discussion of United States v. Chicago Biackhawk Hockey Team, Inc., 468 F.2d 221 (Em.App.1972) and Sampson v. Clark, 162 F.2d 730 (Em.App.1947). Id. at 239-241.
. See also Longview Refining Co. v. Shore, 554 F.2d 1006, (Em. App. 1977), where this court noted that, “Fundamental fairness requires that the regulations be clear so that men of common intelligence need not guess at the meaning and differ as to the application”, citing Boyce Motor Lines v. United States, 342 U.S. 337, 72 S.Ct. 329, 96 L.Ed. 367 (1952). Id. at 1014 n. 20.
. There is some dispute as to whether there was any change in economic conditions during the relevant period which would make prospective application of the rule undesirable, but support its retroactive application. While we do not consider a determination of this question necessary in these cases, we note that there is little, if anything, in the record to support the FEA’s claim of any substantial change between January 1, 1975 and February 1, 1976. Counsel for the FEA suggested in oral argument that this court might take judicial notice of a change in conditions, but in further argument counsel referred primarily to changes between 1973 and 1976, rather than during the relevant period.
. From counsel’s oral argument in response to a question from the court. Counsel stated further that if this court should decide that summary judgment was not appropriate, the FEA would like to prove that the refiners knew or should have known from the language of the regulations that proportional recovery was not correct, that the FEA would like to see the internal communications which indicated the refiners’ compliance strategy and compliance decisions, and that the refiners either deliberately or negligently refrained from seeking official clarification when they realized the auditors were wrong.
. See Standard Oil I, 440 F.Supp. at 372; Phillips Pet. I, 435 F.Supp. at 1249.
. These statements were made in part in support of proposals within the FEA for rulemak-ing or class exception.
. It is undisputed that many of the refiners relied on interpretations and instructions from FEA officials and auditors. This obviously had some bearing on the questions of whether the regulations required an NPCI Last method of recovery. It was not necessary, however, to establish a reliance by all of the refiners to support the courts’ conclusion that there was no required sequence of recovery rule.
. Nor does the fact that a substantial amount of money may be involved preclude summary judgment where there is no genuine issue of fact material to the decision and the case is *1067decided solely on legal issues. Moreover, while the FEA has estimated that as much as $1.3 billion may be involved, there is no evidence to establish any specific amount. As noted supra, the FEA recognized that a NPCI Last sequency of recovery rule would have caused refiners to reduce inventories and cut back domestic refinery output, and would also have caused wide monthly fluctuations in prices. This is made clear in a written statement prepared for presentation to Congressman Dingell’s Subcommittee on Energy and Power of the Committee on Interstate and Foreign Commerce, by Gor-man Smith, the Assistant Administrator of the FEA. Smith stated that the $1.3 billion figure was the “theoretical” aggregate difference between the NPCI Last and proportional methods of recovery. He explained:
It is important to keep in mind that throughout the period in question, market forces rather than FEA regulations were determining prices. Actual prices were well below, in most cases, FEA regulated maximum lawful levels. In fact, the total amount of unrecovered increased product costs for the companies concerned during this period turn out to be of about the same magnitude as the $1.3 billion difference between the two methods. The fact that companies used a sequence of recovery other than the one intended may, therefore, have made no difference in actual product prices charged to consumers.
(R. 2200, 2208-09). The amount could only be determined by an analysis on a company by company basis.
. The information which the FEA seeks to develop on remand would be relevant and crucial in resolving a good faith reliance claim, but not in determining the validity of the FEA’s present interpretation of the regulations.
The question of granting exception relief for “serious hardship or gross inequity” would not be before the district courts on remand. Both courts dismissed the allegations raising factual issues relating to good faith reliance on representations of FEA officials. The Ohio court expressly recognized that the exhaustion doctrine with respect to exception relief (see 10 C.F.R. §§ 205.58 and 205.100-205.109 (1978)) would be applicable to those issues, which it held were not ripe for judicial review. 440 F.Supp. at 370-371. As far as the record discloses, the claims for exception relief are still pending before the Agency.
Moreover, remand to the FEA for consideration of individual exception relief for “serious hardship or gross inequity” in disregard of the controlling point here — the invalidity in the absence of rule-making of the Agency’s NPCI Last interpretation — would be pointless. As the Delaware court recognized, a ruling on the legal issues adverse to the FEA obviates the need for the refiners to participate in the exception proceedings. Phillips Pet. I., 435 F.Supp. at 1248.
. It is true that the Delaware court in holding that the NPCI Last rule was not entitled to retroactive application referred to the “reliance on the absence of a required NPCI Last method of recovery” and the “significant burden a retroactive imposition” of the rule would have on the refiners. Phillips Pet. II, 449 F.Supp. at 748. It does not appear, however, that in reaching these conclusions the court relied on disputed facts. The extent of good faith reliance of the refiners on the FEA’s interpretations is, of course, in dispute, but is not here in issue.
. Evanson v. Union Oil Co., CA. No. 4—75-Civ. 671 (D.Minn.).
. The memorandum stated, inter alia, that the petroleum pricing regulations were not “clear as to the inter-relationship of product cost recovery and non-product cost recovery”, and recognized two possible interpretations:
1. Non-product costs cannot be recovered until all product costs are recovered. This interpretation is supported by a literal reading of the regulations.
2. Some product costs can be held “in the bank” and non-product costs recovered instead. This interpretation is supported by the actual administration of the regulations by the FEA and by their report forms.
It recommended “switching] signals” on cost recovery to recover NPCI first, and concluded “[w]e’ll test our interpretation of the rules and, if we’re wrong, we’ll have lost nothing”. Appellant’s Supplemental Brief (accompanying Appellant’s Motion for Leave to File Supplemental Brief), Attachment A, p. 2-3.
. Appellant’s Supplemental Brief (accompanying Appellant’s Motion for Leave to File Supplemental Brief), p. 2.
. Id. at 10.
. Appellees’ Opposition To The Government’s Motion For Leave To File Supplemental Brief, App. A., p. 13.