The Navy appeals from a district court’s permanent injunction prohibiting disclosure of a subcontractor’s technical data. We affirm the injunction, holding that under any applicable regulations, the subcontractor never surrendered disclosure rights to the Navy.

The challenged injunction was obtained by the appellee Dowty Decoto, a manufacturer of aeronautical equipment. Since 1971 Dowty Decoto has supplied the Navy with “repeatable holdback bars” used in launching F-14 Tomcat fighter planes from aircraft carrier decks. Decoto has supplied the bars pursuant to a subcontract with Grumman Aerospace Corp., the prime contractor supplying the Navy with F-14s. Decoto also sells the bars directly to the Navy on a purchase order basis for use as spares. In addition to the F-14, Decoto also supplies holdback bars for the F-18 Hornet and T-45A trainer aircraft.

On all drawings and data Decoto supplied pursuant to the subcontract, Decoto placed a restrictive legend stating that the data was proprietary and subject only to limited disclosure rights under the contract. It is not disputed that the form of the legend was appropriate for reserving limited disclosure rights in Decoto.

In 1983 the Navy wrote to Decoto asking Decoto voluntarily to remove the restrictive legends from data it had furnished the Navy. Decoto refused, stating that the Navy had never obtained disclosure rights from Decoto. Three years later, the Navy requested Decoto to substantiate its position that the government had acquired only limited rights in the data. After an informal administrative review of Decoto’s submissions, and some informal discussions, the Navy handed down an administrative decision in a letter dated April 27, 1987, advising that Decoto had failed to substantiate its use of restrictive rights legends. It advised that it would obliterate or ignore the legends on the data, and would disclose the data to third parties for the purpose of obtaining competitive bids. Decoto then filed this suit for a permanent injunction in district court, pursuant to the Administra tive Procedure Act (“APA”), 5 U.S.C. § 706 (1982), to prohibit the Navy from disclosing the data. The district court granted the injunction.

There is no dispute that unless the Navy has a right to Decoto’s data and drawings, they otherwise represent trade secrets of Decoto. The Trade Secrets Act forbids government agents from disclosing confidential information “in any manner or to any extent not authorized by law.” 18 U.S.C. § 1905 (1982). If the Navy has no authority to disclose the holdback bar data, its disclosure of Decoto's trade secret would violate section 1905, and “any disclosure that violates § 1905 is ‘not in accordance with law’ within the meaning of 5 U.S.C. § 706 (2)(A).” Chrysler Corp. v. Brown, 441 U.S. 281, 318 , 99 S.Ct. 1705, 1726 , 60 L.Ed.2d 208 (1979). Thus, the APA authorizes this injunction preventing the Navy from disclosing Decoto’s data, provided that such disclosure violates the Trade Secrets Act. Id. at 316-17 , 99 S.Ct. at 1724-25 ; Conax Florida Corp. v. United States, 824 F.2d 1124, 1128 (D.C.Cir.1987).

Our determination of whether the Navy’s action was properly enjoined as a violation of the Trade Secrets Act is in turn guided by regulations governing the Navy’s authority to disclose the data in the absence of Decoto’s acquiescence. The contentions of the parties center on a particular provision of the Armed Services Procurement Regulations (ASPR), regulations promulgated by the Department of Defense governing the acquisition of items for military use, which were in effect when the contract between Decoto and Grumman was signed.

The provision at issue is contained in ASPR §§ 9-202 & 9-203, 32 C.F.R. §§ 9-202 & 9-203 (1965), which deal with rights in technical data. Section 9-202.2 declares the governmental policy of granting to the government unlimited rights to disclose data concerning any item developed at government expense. The policy restricts governmental disclosure of data only where an item was developed at private expense, and where the contractor takes care to mark all data and drawings with a legend prescribed by the regulations setting forth the proprietary nature of the data and the contract under which the data was furnished. Section 9-203(a) implements the policy by requiring that the text of section 9-203(b), which takes the form of a contract clause, be inserted into all government contracts. The language of section 9-203(b) carries out the apportionment of data rights anticipated by the ASPR.

Throughout the administrative and district court proceedings, as well as in this appeal, the dispute between Decoto and the Navy has centered on two issues, one legal and one factual. The legal issue concerns whether ASPR § 9-203 applies at all between Decoto and the Navy, since the form clause language anticipated by the regulations was never inserted into the Decoto-Grumman subcontract and the Navy was not a party to that contract. The factual issue concerns whether, assuming that the regulations do apply, the holdback bar was “developed” at private expense within the meaning of the regulations.

Regarding the application of the regulations to the Decoto-Grumman subcontract, the government maintains that the regulations have general force of law, and must be read into the provisions of all government prime contracts and subcontracts to control the rights of contractors and subcontractors to data furnished the government. Decoto, on the other hand, argues that the regulations operate only through the inclusion of the appropriate form clause language into particular contracts. It argues that because the relevant ASPR form contract clause from section 9-203(b) was never inserted into its subcontract with Grumman, the government does not have unlimited data disclosure rights under the regulations, and that Decoto’s rights to restrict disclosure of the data are governed by the provisions that explicitly appear in the subcontract. These provisions, affirmatively negotiated between Decoto and Grumman, expressly provide that the normal ASPR data rights provision would not be used because it was “unacceptable to Decoto,” and that Decoto would grant only limited data rights.

The district court agreed with Decoto, holding that in the absence of explicit contractual language granting the Navy unlimited data rights, the only way the Navy could assert any rights under the Decoto-Grumman subcontract was as a third-party beneficiary. As a third-party beneficiary, reasoned the district court, the Navy could assert only the rights granted by the contract to Grumman, and the contract specifically withheld from Grumman the right to disclose technical data. The district court concluded that the Navy contracting officer’s analysis of the ASPR as granting unlimited data rights to the Navy for items not developed at private expense was contrary to law, and thus could not stand. In the alternative it held that Decoto had developed the bar at private expense and that the contracting officer’s decision to the contrary was arbitrary and unsupported.

The government on appeal vigorously contests the ruling on the applicability of the regulations, maintaining that under the line of Court of Claims cases beginning with G.L. Christian & Assocs. v. United States, 312 F.2d 418 (Ct.Cl.), cert. denied, 375 U.S. 954 , 84 S.Ct. 444 , 11 L.Ed.2d 314 (1963), the regulations should be regarded as a part of any government contract or subcontract, regardless of whether the form clause language of the ASPR is inserted into the contract and regardless of whether the government is a party to the contract.

Somewhat surprisingly, the effect of the ASPR on government subcontracts when the form clause text of the ASPR has not been inserted into the contract, and when the government is not a party to the contract, has never been discussed in any court decision. The Navy cites G.L. Christian, 312 F.2d at 424 , and also De Matteo Const. Co. v. United States, 600 F.2d 1384, 1391 (Ct.Cl.1979), for the proposition that in a contract between a government agency and a private party, any clause required by federal regulations will be incorporated into the contract even if the contract fails explicitly to include the clause. Those decisions do not resolve the situation here, however, because both of those cases concern interpretation of direct contracts between the government and its prime contractor. The leading Armed Services Board of Contract Appeals case dealing with the respective rights of the government and a subcontractor in technical data, In re Bell Helicopter Textron, 85-3 B.C.A. (CCH) ¶ 18,415, 1985 WL 17050 (A.S.B.C.A.1985), does not address this particular situation either, because both the prime contract and subcontract in that case explicitly incorporated the ASPR form clause text from section 9-203(b). Id. at 92,370-71, 92,374 . Thus, the government asks us to render a decision of first impression.

While we recognize the significance of the legal issues presented in this regard, we find it unnecessary in this case to resolve them, for the decision of the district court must be affirmed on its alternative, factual grounding. Even if the ASPR regulations were read into the subcontract and superseded its express terms, Decoto nevertheless retained its rights to the technical data because the holdback bar was developed at private expense. See ASPR §§ 9.202.2(c) & 9-203(b) clause (b)(2)(h).

Under the APA, we may overturn the contracting officer's decision that the Navy was authorized to disclose Decoto’s data only if the decision was arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law. See 5 U.S.C. § 706 (2)(A). After reviewing the record, we conclude that it compelled the district court’s holding that the bar was developed at private and not government expense. The Navy contracting officer’s decision was arbitrary, and its implementation was therefore properly enjoined by the district court.

The contracting officer’s decision relied in part upon language of the Decoto-Grum-man subcontract, which the Navy contends created unlimited data rights in the government under the ASPR. The subcontract language relied upon recited that design and development were within the subcontract’s scope. The subcontract language calls for Decoto to “design, develop, manufacture, test and deliver all items as required.”

The Navy, however, is mistaken in its belief that the recitals of a contract alone can determine whether an item was actually developed at private expense. ASPR § 9-203(b) clause (b)(l)(i) purports to grant unlimited rights in the government to “technical data resulting directly from •performance of experimental, developmental, or research work which was specified as an element of performance in this or any other Government contract or subcontract” (emphasis added). The regulation requires actual development and work, not merely contract recitals. Procurement authorities use a test based on physical and economic reality, not language, to determine which party actually “develops” an item within the meaning of the statutes and regulations. This test has now been codified within the new Federal Acquisition Regulations System (FARS). See 48 C.F.R. § 227.471 (1987).

The leading administrative decision in this area, from which the current regulation is derived, is In re Bell Helicopter Textron, 85-3 B.C.A. (CCH) ¶ 18,415 (A.S.B.C.A.1985). It concerned a defense research project that had gone through various phases of funding alternately provided by the government and the private contractor. The Armed Services Board of Contract Appeals there recognized that the crucial factor in determining who “developed” an item concerned who took the risk of investing money to transform the item from a speculative idea into a workable item that would probably succeed in its intended use. The Board defined the term “developed” accordingly:

85-3 B.C.A. (CCH) at 92,434.

The Department of Defense adopted this “workability” definition in a regulation it implemented in 1987 for defining the term “developed” in this context:

48 C.F.R. § 227.471 . The Navy points to no authority adopting a different definition for the term “developed,” nor does it argue that the definition has changed since the time when the holdback bars were developed.

Under this standard, our review of the record must focus on the realities of who invested the money that transformed the holdback bar from an uncertain idea into a workable device for its intended application. The record overwhelmingly shows that Decoto’s money, and not the Navy’s, played this role.

The record reflects that Decoto clearly had the technology in place and had developed the bar to the point of workability even before Decoto entered into the contract with Grumman. By the time Decoto originally approached the Navy with the design for the bar, Decoto already had two patents in place on the “high energy release locking actuator ring,” which forms the heart of the bar’s design. The Navy apparently believed in the feasibility of De-coto’s existing design, inasmuch as the Navy itself referred Decoto to Grumman for further funding. In negotiating the contract with Grumman, Decoto never quoted or asked for any funds for design effort or production tooling. The contract calls for the production of first units of the bar within a very short time; four prepro-duction units were to be delivered within three and a half months, and six production models were to follow within approximately three more months. Decoto’s technology was sufficiently developed to allow De-coto successfully to meet these commitments.

The entire framework of the Decoto-Grumman subcontract operates as a straight parts procurement agreement rather than one for research and development. The contract calls for Decoto to supply holdback bars to the Navy as finished products. It contains no expenditure category for research and development work. The total price paid to Decoto under the contract represents simply the aggregate of individual payments for manufactured bars and supporting documentation. The contract is of the “fixed-price” type, promising payment of a specific price for each unit delivered, rather than a “cost-type” contract, which would reimburse the contractor for whatever expenses it incurred plus adding percentage for the contractor’s profit. Fixed-price contracts like Decoto’s have not normally been used for projects requiring research and development. See Bell Helicopter Textron, 85-3 B.C.A. (CCH) at 92,401.

The contracting officer’s decision pointed to changes occurring in the bar’s design during the course of performance of the subcontract to support the conclusion that the Navy indirectly financed design and development of the bar through payments by Grumman. The only evidence in the record that supports the position that the Navy actually financed any of the bar’s development is a “Subcontractor Change Proposal” (SCP) that Decoto sent to Grumman during the term of the contract. The original preproduction contract, dated December 1970, carried a price of $72,344.88. Roughly two years later, in November 1972, Decoto sent six SCPs to Grumman in response to Grumman’s request that the bars withstand 2,000 successful launch cycles rather than the 700 cycles demonstrated by the preproduction units. The proposals sought increases in the contract price, all of which were to be passed through to the Navy, requesting a total of $141,875.20 in additional payments to Deco-to. Five of the six SCPs concerned small specific changes in the bar's design, and were approved by Grumman for the full amounts requested. The sixth request, upon which the Navy here relies, was characterized by Grumman as a “change in scope.” It was the largest and most general in nature, and requested $106,724.22, of which Grumman approved only $53,000. As a result, the SCPs added only $88,-158.98, bringing the total government expenditures for the bars from $151,721.94 to $239,880.92 as of that time.

The justification provided by Decoto in the sixth SCP for seeking reimbursement of nearly $107,000 refers to a “completely new design” that had been developed by Decoto in response to Grumman’s demands that the bars last longer. The contracting officer’s decision held, and the Navy argues on appeal, that this demonstrates that the Navy paid for development of the bar, the payment flowing through Grumman during the term of the contract.

Despite Decoto’s assertions at the time that $107,000 was necessary, there exists no evidence in the record to show that the money actually paid by the Navy through the SCP “developed” the bar to workability within the definition established by Bell Helicopter Textron and 48 C.F.R. § 227.471 . The record contains nothing to suggest that prior to this SCP the holdback bar had a low probability of success in its intended application, or that the bar obtained a high probability of success only as a result of the funding provided by the SCP. Indeed, since the government provided less than half of the development costs requested in 1972, and in effect provided only partial reimbursement for development that had already taken place, the SCP does not support the government’s position that the Navy financed the crucial research and development.

Other evidence in the record suggests that the bars had achieved workability before any government money was paid to Decoto, and that the changes that the government helped finance during the course of the contract were aimed at increasing performance rather than achieving workability. When the original Deco-to-Grumman contract was amended to reflect the increase in contract price, the additional payment was not placed under a research and development category, but was accounted for under a new heading of “qualification test.”

The Navy itself recognized that the bars manufactured without the design changes covered by the SCP were workable. This is most clearly evidenced by the fact that, although the Navy was aware of the changes wrought by the SCPs, it nevertheless approved the ordering and use of forty-two pre-change design bars for use in launching F-14s from aircraft carriers. The Navy merely assigned a different part number to these pre-change units to keep track of their shorter life span. There is nothing in the record to suggest that any of these pre-change bars ever failed to operate properly. The record does contain evidence that in over 250,000 deck launches using the bar only one possible operational failure has ever been noted.

The government directs us to language in Bell Helicopter Textron, 85-3 B.C.A. (CCH) at 92,423, suggesting that if a contractor receives even partial reimbursement for development costs previously voluntarily expended, the government may receive unlimited data rights. We do not believe such a rule, even if appropriate in some cases, should apply in a situation like this where the contractor could not reasonably have been aware that an application for reimbursement could later lead to total forfeiture of data rights which the contractor had in good faith sought to retain by appropriate legends. Here the government did not give Decoto any notice of its intent to claim data rights until ten years after the SCP was submitted.

The Navy contracting officer’s findings that the key research and development, as defined under the standard of Bell Helicopter Textron and 48 C.F.R. § 227.471 , occurred after the contract had begun and was financed by the government were arbitrary and unsupported by the record, and are therefore insufficient under the APA to support a holding that the holdback bars were developed other than at private expense. Because Decoto’s holdback bar was privately developed and its technical data contained the proper restrictive legend, ASPR §§ 9-202 & 9-203 granted only restricted data rights in the bar to the Navy. These regulations do not authorize the Navy to disclose Deeoto’s technical data. Such disclosure would violate the Trade Secrets Act, and is therefore properly en-joinable under the APA.

The district court’s entry of injunction against the Navy was proper and is AFFIRMED.

. The ASPR have since been integrated into the new Federal Acquisition Regulations System. See 48 Fed.Reg. 42,103 (1983). The old regulations were amended as necessary and recodified into C.F.R. Title 48. . Congress later enacted a specific statute regarding rights in technical data, 10 U.S.C. § 2320 (Supp. V 1987), directing the Secretary of Defense to prescribe regulations defining the interests and rights of the government, contractors, and subcontractors in technical data. The statute to a great degree merely codifies the then-existing ASPR regulations. H.R.Rep. No. 690, 98th Cong. 2d Sess. 15, reprinted in 1984 U.S.Code Cong, and Admin.News 4237, 4246. Section 2320 and the regulations promulgated under it, 48 C.F.R. Ch. 2, Sbpt. 227.4, carry forward generally unchanged the provisions of ASPR §§ 9-202 & 9-203. . Section 9-203(b) provides in pertinent part: (b) Basic Data Clause. RIGHTS IN TECHNICAL DATA (FEB.1965) (a) Definitions. ****** (b) Government Rights. (1) The Governmental shall have unlimited rights in: (1) technical data resulting directly from performance of experimental, developmental or research work which was specified as an element of performance in this or any other Government contract or subcontract; ****** (2) The Government shall have limited rights in: ****** (ii) technical data pertaining to items, components or processes developed at private expense, other than such data as may be included in the data referred to in (b)(l)(i), (iii), (iv), (v), and (vi); provided that each piece of data to which limited rights are to be asserted pursuant to (2)(i) and (ii) above is marked with the [proper] legend in which is inserted the number of the prime contract under which the technical data is to be delivered and the name of the Contractor or subcontractor by whom the technical data was generated. . Because we reach our holding on this ground, we have no reason to reach the issue raised for the first time on appeal by amicus Proprietary Industries Association. Amicus contends, based on Conax Florida, 824 F.2d at 1131 , that a reasonable time limit should be imposed, after which the contracting officer may no longer challenge restrictive legends. . It is unclear from the record whether the 2,000 cycle requirement was part of the original subcontract or was later requested by Grumman or the Navy. . Total expenditures as of the 1972 SCP approvals would include the preproduction contract price, $72,344.88, plus forty-two bars ordered at $1,889.93 each. . The justification provided by Decoto in the sixth SCP reads in full: Following receipt of a purchase order from Grumman to design and produce Repeatable Release Holdback Bars, Decoto began a program determined to provide Grumman and the Navy with a completely successful unit. Following design and fabrication of two different configurations, both approved by Grumman, it became clear to Decoto that the goal was not attainable with these designs. A completely new design was developed and Grumman Engineering agreed that the concept had merit. Decoto management after reviewing the program status showing that all areas had large cost overruns from the previous designs, made the decision to go ahead. All previous hardware was scrapped and machining began again on the new design. Now, after some additional minor set backs the goal is in sight. The Navy will receive a successful Repeatable Release Holdback Bar, something for which large sums have been spent earlier without success. The redesign and rebuilding stages have been expensive, far in excess of the dollar value of the original purchase order. Decoto committed this money without hesitation in the belief that should the effort produce a unique Hold-back Bar, all or most of these funds could be recovered due to the exceptional value of the product to the Navy. Through this change, Decoto is submitting a recap of the additional Decoto money that was spent. It is requested that Grumman review these statements and acknowledge the diligence and determination of Decoto by increasing Purchase Order 7-04384 by this amount.