This is the first appeal from a decision of the FPC acting under section 2.-75 of its regulations, which establishes a procedure for certification of new sales of natural gas “ ‘notwithstanding that the contract rate [might] be in excess of an area ceiling rate established in a prior opinion or order of this Commission.’ ”1 In Moss v. FPC,2 we upheld the basic validity of section 2.75, but we did not define the permissible conditions of its application. In this case, we must consider the character and quantum of proof needed in the certification of contract rates under the optional procedure.
In this proceeding, the FPC considered the rate provisions of contracts between three producers (Belco, Tenneco and Texaco) and the Tennessee Gas Pipeline Co. The contracts provide for the sale of gas produced from wells recently drilled in the offshore Louisiana area. The FPC approved as “just and reasonable” the basic rate of 45 cents per Mcf provided for in each contract as well as certain yearly escalation features.3 The area rate presently applicable for “new gas” is 26 cents per Mcf.4
*278There are serious weaknesses in the Commission’s justification for this nearly sixty percent increase over the area rates, which were established as recently as 1971. These weaknesses appear both in the way in which the Commission calculated the costs of producing the gas and in the weight which the Commission accorded non-cost considerations.
The staff’s cost presentation, on which the Commission relied partially, sought to estimate the current costs of producing new gas. The staff utilized the methodology consistently employed by the Commission in its area rate decisions, combining information from a number of sources to establish the national cost of finding and producing new gas.5 There is some question initially whether reliance on nationwide data is consistent with the “supply project” approach adopted by the Commission in its opinion, an approach whereby it seeks to ensure that “gas consumers are receiving the lowest cost available increment of supply.”6 FPC Chairman Nassikas, dissenting in part, thought that the “supply project approach” required reference to “actual unit costs ... or . individual project costs.” 7 A decision requiring the Commission to rely on individualized cost data, however, would have to be reconciled with this court’s opinion in Moss, which approved, by implication, the Commission’s avowed intent to rely on “ ‘cost findings embodied in our area rate decisions.’ ” 8 And in any event, another more obvious problem with the Commission’s cost analysis makes it unnecessary to reach this issue.
In its final estimate of production costs, the Commission made use of 1971 as a “test year” in determining productivity — the average number of Mcf added to available reserves per each foot drilled. The Commission’s staff followed the practice, well-established in area rate-making proceedings, of using productivity figures averaged over a period of years. The low end of its cost estimate was based on average productivity over the last 15 to 25 years, and the upper limit of its analysis was based on average productivity between 1967 and 1971. Its calculations yielded a cost range of 28 to 36 cents per Mcf.
The Commission adopted the staff’s upper limit as the basis of the lower limit of its estimate.9 The upper limit of the Commission’s estimate (48 cents per Mcf) was based on productivity statistics for 1971 alone. The 1971 productivity figure (379) was substantially lower than the productivity figures which the staff arrived at by its averaging methods (555-600), and it accounts almost entirely for the higher cost estimates by which the Commission was able to approve the contract rates.10
*279The Commission justifies its departure from past practice in selecting the “test year” approach on the ground that 1971 was the year in which the wells producing the contract gas were drilled. As productivity had been generally on the decline in the years prior to 1971, the statistics for that year, .the Commission argues, offered a more accurate estimate of the actual productivity of the new wells.
The superior accuracy of the 1971 figures is brought into question by evidence on the record. First, as Chairman Nassikas summarizes in dissent, “[t]he results of the wells drilled in 1971 will be reflected for the most part in reserves added in subsequent years. . ” 11 Second, there were certain “statistical revisions” made in 1971 which significantly affect productivity rates for that year. These were negative adjustments in the estimated reserve additions carried over from prior years, and had nothing to do with the actual experience of the industry in 1971. They resulted in a reduction of I.1 billion Mcf. If allowance were made for this purely statistical manipulation, productivity for 1971 would approach 500 Mcf rather than the 379 Mcf relied on by the Commission.12
The reasonableness of the Commission’s adoption of the 1971 “test year” is
further undercut by evidence that the productivity of wells in the Southern Louisiana area was about 4.8 times as high as the national average in 1971.13 Thus, in the name of “accuracy” the Commission moved to a lower productivity rate than had been used by its staff, in the face of evidence that an even higher rate may, in fact, have been closer to reality.
The Commission is certainly free to try out new techniques, but it is constrained to show that its departures from established practice are reasonable,14 particularly where, as here, the change is crucial to its decision.15 It has not made that showing on the record in this case.
The Commission also points k various non-cost factors as justification for its decision. These include the contract rates, as negotiated, the prevailing intrastate rates, the cost of importing gas from other sources (e. g., Canadian gas, coal gas), and the “commodity value of natural gas” (based on a comparison of the contract rates with the cost of “substitutable forms of energy in sixteen areas served by Tennessee and its resale customers.”16). The appellants attack the Commission’s reliance on these factors from a number of different directions. In light of the problems *280with the Commission’s cost estimates,, however, perhaps only two rather general points need to be made.
Even after Permian and Mobil Oil, it is doubtful that non-cost factors can sustain a decision by the FPC which is unsupported by sound cost data. In Mobil Oil, for instance, where great deference was paid to non-cost elements in upholding the Commission’s decision, the Court began with the premise that “[appellant’s] attack on the Commission’s evidence of costs is clearly frivolous.”17 And even if non-cost factors could, under certain circumstances, overcome problems in cost analysis of the sort apparent here, these factors are not entitled to overriding weight in the particular circumstances of this case.
Reliance on non-cost factors has been endorsed by the courts primarily in recognition of the need to stimulate new supplies of natural gas in interstate commerce.18 Here, however, the needed supplies are assured. The gas reserves have already been discovered and tapped.19 More importantly, as the wells are on leases in the federal domain, the gas cannot be sold at all without the Commission’s approval.20 Thus, there is no potential diversion to intrastate commerce. This is not to suggest that offshore producers, as in some sense captives of the Commission, are to be confined to the lowest rate constitutionally permissible, but their peculiar status cannot be ignored in determining what weight market factors should be accorded.
One could argue that the approval of the 45 cent rate sought by these companies would in fact augment interstate supplies by encouraging producers to tap reserves in the area on the assumption that their contracts will be accorded similar treatment. As recognized by the court’s opinion in Moss v. FPC,21 it is the purpose of section 2.75 to stimulate new production by offering relief from area rates to producers under individual contracts. At the same time, however, the court cautioned that section 2.75 could not be administered by the Commission “to substitute contract prices .negotiated between producers and pipelines for established just and reasonable rates.”22 Therefore, the Commission cannot be taken as suggesting, by its decision in this proceeding, that contract rates in other eases will necessarily be approved. And the Commission has sought to make this clear:
In insisting that our decision here is to be read only as a decision on the applications before us, we seek to underscore the essential nature of Section 2.75 cases as proceedings which do not carry industrywide consequences. Our decision here establishes rates and conditions of service for three individual sales. It does no more.23
In light of the flaws in the Commission’s cost analysis, and in light also of the limited relevance of supply considerations to this proceeding, we set aside the Commission’s order and remand for a redetermination of the reasonableness of the contract rates.
We have reason to believe that this disposition will be welcomed by the FPC. For the Commission has abandoned the “test year” approach in more recent cases and has relied instead on *281the cost findings embodied in the applicable area rate decision.24 A remand will allow it to reach a result in this case more consistent with its recent approach and more consistent, incidentally, with the approach which it proposed in initially promulgating section 2.75.25
. Moss v. FPC, 164 U.S.App.D.C. 1, 3, 502 F.2d 461, 463 (1974).
. 164 U.S.App.D.C. 1, 502 F.2d 461 (1974).
. FPC Opinion No. 659 (May 30, 1973), reproduced in designated portions of the record in lieu of an appendix, at 4842. [References to portions of the Commission’s opinion, and the concurring and dissenting opinion of Chairman Nassikas, are hereinafter made to pages in the record.]
. Southern Louisiana Area Rate Proceeding, 46 F.P.C. 86 (1971), aff’d sub nom. Placid Oil Co. v. FPC, 417 U.S. 283, 94 S.Ct. 2328, 41 L.Ed.2d 72 (1974).
. See Permian Area Rate Cases, 390 U.S. 747, 761, 88 S.Ct. 1344, 20 L.Ed.2d 312 (1968).
. R. 4849.
. R. 4911. The companies were given the opportunity to put forth specific cost data on the producing properties involved, but no such data was produced. See Brief for Respondent Federal Power Commission at 27 n. 35.
. Moss v. FPC, supra note 2, at 466, quoting FPC Order No. 455.
. R. 4869.
. In following through with t'e “test year” approach, the Commission also relied on 1971 cost data. Some elements of the 1971 figures, such as drilling costs and dry hole costs, showed increases over prior years, see R. 4873; other elements, such as the cost of producing wells, lease acquisition costs and exploratory overhead, showed decreases, see id. The net effect of the Commission’s use of this data was therefore not great.
The Commission’s implementation of the “test year” approach was not complete. It relied on t'..e cost findings contained in its Southern Louisiana area rate decision with regard to production expenses, regulatory expenses and net liquid credit. See R. 4875. The Commission cites lack of evidence for trending these costs as a justification for falling back on its Opinion 598 findings, although it does not indicate that it sought to develop such evidence on the record.
. R. 4911.
. R. 4912. This is Chairman Nassikas’ calculation. In oral argument, counsel for the FPC disputed this projection, claiming that if the negative revisions were compensated for, the result would be a productivity rate of 426, which would still be sufficient to justify the 450 rate on a cost basis.
. Exhibit No. 31, R. 3560-61.
The Commission’s brief on appeal acknowledges the higher productivity of offshore wells but suggests at the same time that this higher productivity is offset by the higher costs of offshore drilling, which in 1971 averaged 2.5 times more than the costs of onshore drilling. Id. Although testimony on the record cites a “general offsetting effect,” R. 1771, it does not support t' e Commission’s assertion that one factor effectively cancels the other. By simple arithmetic calculation, the return on a dollar spent in 1971 in the offshore area exceeded return on the same dollar spent onshore by a factor of nearly two, if the statistics of Exhibit No. 31 are accepted.
. Cf. City of Detroit v. FPC, 97 U.S.App.D.C. 260, 268, 230 F.2d 810, 818 (1955), cert. denied, 352 U.S. 829, 77 S.Ct. 34, 1 L. Ed.2d 48 (1956) (“if the Commission is now to abandon the treatment historically accorded pipeline-produced gas in rate making on the ground that the ultimate public interest will be better served thereby, the Commission should justify it on the record.”)
. See Permian Area Rate Cases, supra note 5, 390 U.S. at 790-792, 88 S.Ct. at 1372-1373.
. R. 4881.
. Mobil Oil Corp. v. FPC, 417 U.S. 283, 314, 94 S.Ct. 2328, 2349 (1974).
. See, e. g., id. at 314-321, 94 S.Ct. at 2349-2352.
. In this respect, section 2.75 proceedings will all be similar, and distinguishable from area rate proceedings, in which the Commission’s order applies not only to flowing gas but to gas which has yet to be introduced into the market.
. R. 4917.
. 164 U.S.App.D.C. at 3-5, 502 F.2d at 463-465.
. Id. at 4, 502 F.2d at 464.
. R. 4850.
. See, e. g., In re Stingray Pipeline Company, Docket No. CP 73-27 (FPC May 26, 1974).
. We will, absent a showing of special circumstances, accept as conclusive the cost findings embodied in our area rate decisions, as such may be supplemented from time to time by appropriate Commission order. FPC Order No. 455 (Aug. 3, 1973) (emphasis added). As mentioned, see note 7 supra, there were no “special circumstances” asserted by the companies in this case that would have justified a departure from the cost findings on which the Commission’s Southern Louisiana area rates are based.