These are consolidated wrongful death actions arising out of two unrelated crashes of RA-5C naval aircraft in the waters off the coast of Florida. The widows of the two Navy pilots killed in the crashes seek damages from Rockwell International Corp. (“Rockwell”), the manufacturer of the RA-5C aircraft and its ejection system. The district court held that Rockwell was liable for the pilots’ deaths because of defects in the aircraft’s ejection system. Both the widows and Rockwell have appealed.
In this case we confront the question under what circumstances, if any, the doctrine of strict liability in tort, as set forth in section 402A of the Second Restatement of Torts, should be extended to cover manufacturers of military equipment that proves to be defective in design and injures members of the armed forces who are on active duty. We also must address the question whether, under the circumstances of this case, sections 388 and 389 of the Second Restatement impose liability on Rockwell. For the reasons set forth below we reverse the judgment of the district court and remand for further proceedings.
I.
FACTS
Rockwell, under contract with the United States Navy, began development in the mid-1950s of an aircraft capable of sustained flight at altitudes of up to 75,000 feet, and with a potential speed of two and a half times the speed of sound. In the early 1960s, the Navy decided to redesign the aircraft as a supersonic carrier-based reconnaissance aircraft, designated the RA-5C “Vigilante.” The RA-5C was put into use by the Navy in 1962 and was used extensively in Vietnam.
Both the RA-5C aircraft involved in the accidents out of which these cases arise were equipped with the HS-1A escape system. This system was a modified version of an earlier escape system in use in the RA-5C aircraft. The HS-1A system operated by physically restraining the crew in their seats, and then ejecting them ballistically into the airstream by means of a rocket thrust. After ejection, a drogue chute would initiate the opening of a 28-foot parachute to enable the crewmen to descend safely to the ground.
On March 5, 1974, Navy Lieutenant Frank Carson was killed during a daytime training mission when the RA-5C aircraft he piloted caught fire and he was forced to eject from the aircraft. Navy Lieutenant Commander Malcolm McKay was killed on August 13,1974, after ejecting from a burning RA-5C aircraft during a night training mission. Autopsies of the two pilots revealed that their deaths were probably caused by injuries sustained during ejection.
Plaintiffs filed civil actions in the United States District Court for the Central Dis*447trict of California, seeking recovery of damages for the death of plaintiffs’ decedents under theories of negligence, breach of warranty, and wrongful death.
The cases were consolidated for trial and, after an evidentiary hearing, the district court determined that it had admiralty jurisdiction over the actions pursuant to the Death on the High Seas Act, 46 U.S.C. §§ 761-767.1 The district court found that Rockwell properly was liable for the design of the HS-1A escape system under the principles of tort law set forth in sections 388, 389, and 402A of the Second Restatement of Torts. The court declined to impose liability under these principles for the design of the RA-5C aircraft. It entered judgment after a trial on the merits in favor of plaintiff Carson for $385,703.00 and in favor of plaintiff McKay for $325,-850.00. Carson and McKay seek review of the measure and amount of damages awarded in their respective judgments. Rockwell also appeals, contending that military suppliers should not be liable to servicemen for injuries caused by defects in military hardware.2 Our disposition of Rockwell’s appeal makes it unnecessary to address the appeals of Carson and McKay.
II.
LIABILITY UNDER SECTION 402A OF THE SECOND RESTATEMENT OF TORTS
The district court, as stated above, held that Rockwell was liable under section 402A of the Second Restatement of Torts for defects in the design of the HS-1A escape system.3 We applied the principles of this section in admiralty in Pan-Alaska Fisheries, Inc. v. Marine Construction & Design Co., 565 F.2d 1129 (9th Cir.1978). But in Pan-Alaska we did not hold that strict liability applies for all purposes and for all defendants.4 Section 402A is not a federal statute. It should be applied only when the purposes it seeks to serve dictate its application. When that is not the case it has no independent force. To apply it merely because it is there is to abdicate judicial responsibility.
Mindful of this responsibility, we conclude that only under the limited circumstances we shall enumerate below should a manufacturer be held strictly liable in tort for injuries to a serviceman on active duty caused by design defects in military equipment.
*448A. Feres-Stencel Doctrine
We commence our analysis with Feres v. United States, 340 U.S. 135, 71 S.Ct. 153, 95 L.Ed. 152 (1950). In that case the Supreme Court held that the United States is not subject to liability under the Federal Tort Claims Act, 28 U.S.C. § 2674, to a member of the armed forces who sustains an injury while on active duty. The scope of governmental immunity was broadened recently in Stencel Aero Engineering Corp. v. United States, 431 U.S. 666, 97 S.Ct. 2054, 52 L.Ed.2d 665 (1977). There, the Court held that the Federal Tort Claims Act precludes the United States from indemnifying a third party for damages paid by it to a member of the armed forces who is injured during military service. The Stencel Court explained that allowing indemnity would subject the United States to varying degrees of liability, depending on the situs of the accident, would require the United States to pay indirectly to the serviceman what the Veterans’ Benefits Act forbids it to pay directly, and would interfere with military discipline. Id. at 672-73, 97 S.Ct. at 2058-59.
Thus, under the circumstances of these cases, the United States would be immune both from direct tort liability as well as from the obligation of indemnifying Rockwell for damages it might be required to pay.
B. Government Contractor Defense
Given the immunities of the United States in cases such as these, the question arises whether a supplier of military equipment should be required to shoulder directly and immediately the entire burden of the liability to an injured serviceman. Some courts, when confronted with this issue, have relied on the so-called government contractor defense.5 This rule, first articulated by the Supreme Court in Yearsley v. W.A. Ross Construction Co., 309 U.S. 18, 60 S.Ct. 413, 84 L.Ed. 554 (1940), protects a government contractor from liability for acts done by him while complying with government specifications during execution of performance of a contract with the United States. See Myers v. United States, 323 F.2d 580, 583 (9th Cir.1963). The rule has been applied when the United States is immune from suit. Dolphin Gardens, Inc. v. United States, 243 F.Supp. 824, 827 (D.Conn.1965).
While the government contractor defense covered at first only construction projects, it has recently been applied by several courts to military equipment design defect cases.6 See Note, 23 B.C.L.Rev. 1025, 1055-64 (1982). For example, in Sanner v. Ford Motor Co., 144 N.J.Super. 1, 364 A.2d 43 (1976), aff’d, 154 N.J.Super. 407, 381 A.2d 805 (1977), cert. denied, 75 N.J. 616, 384 A.2d 846 (1978), the court held that when a manufacturer produces a jeep in compliance with government specifications, the manufacturer cannot be held strictly liable for defects in the government’s design specifications. Similarly, in Casabianca v. Casabianca, 104 Misc.2d 348, 428 N.Y.S.2d 400 (1980), the manufacturer of kitchen equipment made for the Army and in accordance with Army specifications was held not to be subject to liability for defects in the equipment. Finally, in In Re Agent Orange Product Liability Litigation, 534 F.Supp. 1046 (E.D.N.Y.1982), the court approved a government contractor defense for manufacturers of a chemical defoliant where the government set or ratified performance specifications for a product, the manufac*449turer met those specifications, and warned the government of known dangers from using the product. Id. at 1055. See also Littlehale v. E.I. DuPont de Nemours & Co., 268 F.Supp. 791 (S.D.N.Y.1966), aff’d, 380 F.2d 274 (2d Cir.1967) (no duty to print warnings on blasting caps where not required by Navy specifications).
The reasons for applying the government contractor defense to suppliers of military equipment with design defects approved by the government parallel those supporting the Feres-Stencel doctrine. First, the Supreme Court emphasized in Stencel that the United States cannot be directly or indirectly liable to servicemen injured by defective military products.7 But holding the supplier liable in government contractor cases without regard to the extent of government involvement in fixing the product’s design and specifications would subvert the Feres-Stencel rule since military suppliers, despite the government’s immunity, would pass the cost of accidents off to the United States through cost overrun provisions in equipment contracts, through reflecting the price of liability insurance in the contracts, or through higher prices in later equipment sales. See In Re Agent Orange Product Liability Litigation, 506 F.Supp. 762, 793-94 (E.D.N.Y.1980), rev’d on other grounds, 635 F.2d 987 (2d Cir.1980); Dolphin Gardens, Inc. v. United States, 243 F.Supp. 824, 827 (D.Conn.1965); Note, 55 N.Y.U.L.Rev. 601, 618 (1980). As the Court explained in Stencel:
To permit [petitioner] to proceed .. . here would be to judicially admit at the back door that which has been legislatively turned away at the front door. We do not believe that the [Federal Tort Claims] Act permits such a result.
432 U.S. at 673, 97 S.Ct. at 2059 (quoting from Laird v. Nelms, 406 U.S. 797, 802, 92 S.Ct. 1899, 1902, 32 L.Ed.2d 499 (1972)).
Second, to hold military suppliers liable for defective designs where the United States set or approved the design specifications would thrust the judiciary into the making of military decisions. Although judges must decide cases arising from fields of endeavor of which they know little, their otherwise omnicompetence confronts its limits in military matters. At this point, it must be acknowledged, separation of powers becomes a proper concern. See In Re Agent Orange Product Liability Litigation, 534 F.Supp. at 1054; Montgomery v. Goodyear Tire & Rubber Co., 231 F.Supp. 447, 450 (S.D.N.Y.1964); see also Morrison v. Larsen, 446 F.2d 250, 253 (9th Cir.1971); Aero Corp. v. Department of the Navy, 493 F.Supp. 558, 567 (D.D.C.1981). Trials on design defects where government specifications are at issue would “involve second-guessing military orders, and would often require members of the Armed Services to testify in court as to each other’s decisions and actions.” Stencel, 431 U.S. at 673, 97 S.Ct. at 2059. These trials would raise concerns about their effect on military discipline, id., as well as on national security.8
Third, it should be noted that in setting specifications for military equipment, the United States is required by the exigencies *450of our defense effort to push technology towards its limits and thereby to incur risks beyond those that would be acceptable for ordinary consumer goods. A supplier is frequently unable to negotiate with the United States to eliminate those risks. As one court put it:
Where, as here, manufacturers claim to have been compelled by federal law to produce a weapon of war without ability to negotiate specifications, contract prices or terms, the potential for unfairly imposing liability becomes great. Without the government contract defense a manufacturer capable of producing military goods for government use would face the untenable position of choosing between severe penalties for failing to supply products necessary to conduct a war, and producing what the government requires but at a contract price that makes no provision for the need to insure against potential liability for design flaws in the government’s plans.
In Re Agent Orange Product Liability Litigation, 506 F.Supp. at 794. See generally Tobak, A Case of Mistaken Liability: The Government Contractor’s Liability for Injuries Incurred by Members of the Armed Forces, 13 Pub.Cont.L.J. 74 (1982); Note, 23 B.C.L.Rev. 1025 (1982).
Finally, a government contractor defense provides incentives for suppliers of military equipment to work closely with and to consult the military authorities in the development and testing of equipment. The defense therefore encourages fixing the locus of responsibility for military equipment design with more precision than is possible under a system where the government contractor rule is not allowed.
While conceding that the government contractor defense can apply to military products, plaintiffs-appellants argue that it should not apply to these cases. As they see it, the defense is only available where the specifications in the contract leave no discretion to the supplier in the formulation of the product’s design. They rely on O’Keefe v. Boeing Co., 335 F.Supp. 1104 (S.D.N.Y.1971), and on Merritt, Chapman & Scott Corp. v. Guy F. Atkinson Co., 295 F.2d 14 (9th Cir.1961), to support this assertion.9 Plaintiffs-appellants’ reliance is misplaced.
In O’Keefe, the court rejected Boeing’s claim that it was not liable for an alleged defect in a B-52 because of the involvement of the United States in the development of the aircraft. 335 F.Supp. at 1122-24. But the discussion of the design discretion issue in O’Keefe was dictum, since the court found that the design was not defective. Moreover, the O’Keefe dictum is not compatible with the later Stencel holding.
Notwithstanding its insistence that “compulsion” is an essential element of the government contractor defense, Merritt, Chapman is distinguishable. It held that a subcontractor who built a faulty cofferdam in a dam project' can be held liable for damage caused by the collapse of the cofferdam. The government contractor claim was rejected because the contract with the United States left the design, materials, and method of construction entirely to the discretion of the subcontractor. The United States merely set the height requirement for the cofferdam. Id. at 15-16.
Under these circumstances, Merritt, Chapman properly precludes the government contractor rule. When only minimal or very general requirements are set for the contractor by the United States the rule is inapplicable. The situation is different where the United States reviewed and approved a detailed set of specifications. This is precisely what may have happened in the present cases. See R.T. at 1303-04, 1309-10, 1486-87.
It is at this point that the Feres-Stencel doctrine comes sharply into focus. The United States has limited its liability to service personnel injured while on active duty. It is consistent with this limitation to construe the government contractor rule so as to avoid imposing on the contractor liability properly attributable to acts of *451government: The narrower the scope of this rule the greater the extent of this “misplaced” liability. While to narrow the rule aids members of the armed services and their dependents, it also imposes indirectly burdens on taxpayers and/or inflation bearers which the Feres-Stencel doctrine precludes. It follows that the scope of the government contractor rule, when applied in cases involving military personnel, should be drawn somewhat more broadly than the dictum in Merritt, Chapman might suggest.
To summarize, we hold that under the Feres-Stencel doctrine and the government contractor rule, a supplier of military equipment is not subject to section 402A liability for a design defect where: (1) the United States is immune from liability under Feres and Stencel, (2) the supplier proves that the United States established, or approved, reasonably precise specifications for the allegedly defective military equipment, (3) the equipment conformed to those specifications, and (4) the supplier warned the United States about patent errors in the government’s specifications or about dangers involved in the use of the equipment that were known to the supplier but not to the United States. The imposition of this duty to warn of known defects is necessary to enable the United States to balance the risks and benefits inherent in the use of the equipment. Cf. In Re Agent Orange Product Liability Litigation, 534 F.Supp. at 1055; Note, 23 B.C.L.Rev. at 1074-85.
We recognize that the term “military equipment” is somewhat imprecise, and that at some point lines will have to be drawn. We need not do so here. The line, however, lies somewhere between an ordinary consumer product purchased by the armed forces — a can of beans, for example — and the escape system of a Navy RA-5C reconnaissance aircraft. The latter falls within the term while the former does not.
We also note that the rule enunciated here does not relieve suppliers of military equipment of liability for defects in the manufacture of that equipment. To hold otherwise would remove the incentive from manufacturers to use all cost-justified means to conform to government specifications in the manufacture of military equipment.
C. Policy Considerations in the Imposition of Strict Liability
The reasons for imposing strict liability as set forth in section 402A are inapplicable when the elements of our holding exist. See Note, 23 B.C.L.Rev. at 1080-85. Courts and commentators have identified four principal reasons for imposing strict liability on an accident producing activity — enterprise liability, market deterrence, compensation, and implied representation of safety.10 See Note, 33 Stan.L.Rev. 535, 536 & n. 7 (1981). We shall consider each.
1. Enterprise Liability
Under the enterprise liability rationale, when a product’s price reflects the cost of accidents caused by the use of the product, that price will rise. Increased prices will then discourage consumers from purchasing risky products, and thereby lower accident costs to society. See Klemme, The Enterprise Liability Theory of Torts, 47 U.Colo.L.*452Rev. 153, 158 (1976). Cf. Pan-Alaska Fisheries, 565 F.2d at 1135.
However, the rationale rests on two assumptions. These are that consumers underestimate the risks involved in a product’s use, and will therefore overconsume the product unless the product’s price reflects the cost of accidents, and that demand for a product is elastic — that is, that it will decrease as the product’s price rises. Note, 33 Stan.L.Rev. at 537 & n. 8.
Neither of these assumptions applies in the usual case to sales of military equipment to the government. First, the armed forces are aware of most, although sometimes not all, the risks involved in using military equipment. They undertake a constant program of testing and evaluating such equipment. Higher prices would not affect significantly their awareness of the safety risks involved in the use of the equipment. In addition, within broad limits demand is not elastic for military equipment. Rather, government purchases of military equipment are planned in advance, and are based on considerations of military and political strategy, as well as on the government’s assessment of the risks and benefits involved in the use of the equipment. Thus, including the cost of accidents in the price of sales to the military would probably have little or no effect on product sales. Meeting adequately the needs of national defense, not accident costs, is the ultimate standard by which purchases of military equipment must be measured.
2. Market Deterrence
A second reason for imposing strict liability is to deter manufacturers from marketing unsafe products by encouraging the use of cost-justified safety features. See W. Prosser, The Law of Torts § 4, at 23 (4th ed. 1971). The safer the product, the argument runs, the lower the cost of accidents. This should reduce the product’s price which, in turn, should increase the sales of the product.
But in the case of military equipment, as noted above, the demand for such equipment is quite inelastic. Moreover, the government, the sole purchaser of most military equipment, has both the ability to recognize safety problems in military equipment and to negotiate with suppliers to remedy those problems. It constantly balances the safety of the article against the imperatives of national defense. Strict liability would no doubt increase defense costs but would do little not already being done to increase the use of safety features in military equipment. See In Re Agent Orange Product Liability Litigation, 506 F.Supp. at 793. Increased defense costs, on the other hand, will diminish either other expenditures, public or private, or the level of national defense, if the level of total expenditures for that purpose were to be held constant.
3. Compensation
A third justification for strict liability is that it provides compensation for victims of accidents caused by defective products. Restatement (Second) of Torts § 402A, comment c. In the case of injured military personnel, however, the Veterans’ Benefits Act provides what the Supreme Court called “a generous military compensation scheme,” and “a swift, efficient remedy.” Stencel, 431 U.S. at 672-73, 97 S.Ct. at 2058-59.11 Thus, the serviceman or his family will not go uncompensated, unlike the case of an ordinary consumer injured by a defective product. It is true, of course, that strict liability would increase that compensation, but it can hardly be said that any such increase was anticipated at the time of enlistment.
4. Implied Representation
Finally, it has been reasoned that by marketing a product, a supplier makes an im*453plied representation that the product, if put to its intended use, will not be unreasonably dangerous and will meet the safety standard expected of similar products. If the product proves to be defective, consumers should receive compensation for the disappointment of their reasonable expectations of safety. Note, 33 Stan.L.Rev. at 544; Restatement (Second) of Torts § 402A, comment i.
Members of the armed forces are not ordinary consumers with respect to military equipment. Their “reasonable expectations of safety” are much lower than those of ordinary consumers. They recognize when they join the armed forces that they may be exposed to grave risks of danger, such as having to bail out of a disabled aircraft. This is part of the job. The Nation sometimes demands their very lives. This is an immutable feature of their calling. To regard them as ordinary consumers would demean and dishonor the high station in public esteem to which, because of their exposure to danger, they are justly entitled.
D. Application of Our Holding to the Facts of These Cases
The application of our holding to the facts of these cases requires that we reverse and remand the judgments below. It is clear, and we so hold, that in these cases the United States is immune from liability for the design of the HS-1A ejection system under Feres and Stencel. Moreover, the district court held that the defect in the system was its design, not its failure to conform to government specifications. Finally, there is no allegation that Rockwell failed to warn the United States of dangers known to Rockwell but not to the Navy.
The present record, however, does not permit us to say with assurance that the United States set or approved reasonably detailed specifications for the HS-1A system. On the one hand, Rockwell alleges that the United States was deeply involved in the process of designing and approving the system. On the other hand, plaintiffs-appellants contend that the United States did little more than send Rockwell a letter asking them to come up with a new ejection system and agree to purchase Rockwell’s completed design.12
We remand these cases to the district court to determine whether the United States set or approved reasonably detailed specifications for the HS-1A ejection system. If the district court finds that the involvement of the United States was limited in the manner the plaintiffs-appellants assert — that is, if the United States neither set specifications for the system (other than general outlines of what type of system it required) nor approved Rockwell’s final reasonably detailed specifications (by examining and agreeing to a detailed description of the workings of the system) — then Rockwell is subject to strict liability under.the rule set forth in section 402A. We note that Rockwell, the supplier, has the burden of proving by a preponderance of the evidence that the United States established, or approved, reasonably precise specifications for the ejection system.
III.
LIABILITY UNDER SECTIONS 388 AND 389 OF THE SECOND RESTATEMENT OF TORTS
The district court also found that Rockwell was liable under Restatement (Second) of Torts §§ 388 and 389 for breach of a duty to “test, measure or evaluate the dynamics of the ejection process on the head, helmet, and neck of the ejecting crewman,” and to withdraw the HS-1A system after “continued use [showed] ... that a pattern of neck injuries was developing.” We reverse.
We commence by observing that this court has not yet adopted sections 388 and 389 as a basis for liability in admiralty in this circuit. But even if we had, these sections would not support the district *454court’s holding. First, many of the arguments discussed in Part II apply with equal force here. For example, the high incidence of government evaluation of the safety of military equipment places it in a position quite different from that of an ordinary purchaser or consumer. Also the socio-economic justifications for strict liability, as set forth above in Part II-C, continue to be inapposite when liability under these sections is considered. In addition, as explained below, the requirements for liability under sections 388 and 389 have not been met.
A. Section 388
Under section 388, a supplier of a “dangerous chattel” is liable to those whom the supplier should expect to use the chattel if (1) the supplier knows or has reason to know that the chattel is or is likely to be dangerous; and (2) it has no reason to believe that users will realize the danger; and (3) it fails to warn the users of the chattel’s dangerous condition.13 All three criteria must be satisfied for liability to attach under section 388. Dougherty v. Hooker Chemical Corp., 540 F.2d 174, 178 (3d Cir.1976). Here both the second and third requirements for liability are lacking.
The Navy, the principal user of the HS-1A system, was aware of any injuries incurred while using the system. It “realized the danger”; it was engaged in a continuous process of evaluating the system, and of exchanging information on its performance with Rockwell. See, e.g., R.T. at 1342-44. Thus, there was no reason for Rockwell to believe that the Navy was unaware of problems with the HS-1A system, and Rockwell therefore had no duty to warn the Navy about the system’s “dangerous condition.”14 See section 388, comment k; see also Strong v. E.I. DuPont de Nemours Co., 667 F.2d 682, 687 (8th Cir.1981).
In reaching its conclusion the district court relied on the following assumptions: Rockwell had a duty to test the HS-1A system; Rockwell failed adequately to test the system; testing would have revealed that the system was dangerous; Rockwell would then have been obliged to warn the Navy about these dangers and to withdraw the system from the market; and warning and withdrawal would have prevented the accidents. These assumptions are flawed.
First, it is questionable whether section 388 imposes a duty on a supplier to test a product for latent defects. See Lockett v. General Electric Company, 376 F.Supp. 1201, 1208 (E.D.Pa.1974), aff’d mem., 511 F.2d 1393 (3d Cir.1975); see also Sears, Roebuck & Co. v. Marhenke, 121 F.2d 598, 600 (9th Cir.1941). To impose on Rockwell a duty to test for latent defects would cause it to become a virtual guarantor of the proper performance by the Navy of its duties. And neither the text nor the comments to section 388 indicate that there is a duty under that section on the part of a supplier to withdraw a product from the hands of the user, particularly when that user is the Navy of the United States.
Second, even if there were such duties, the record does not show that breach of the duties proximately caused the injuries at *455issue in this case. There is no evidence that the Navy would have removed the system from its existing RA-5C aircraft had the Navy been aware of the possibility of the type of accident with which we are concerned. Indeed, the Navy evaluated the system subsequent to the accidents and found, on balance, that it was “safe-for-flight.” Exhibit 112 (United States Navy, A-4, A-5, & A-6 Aircraft Escape System Neck Injury Investigation, Nov. 12, 1974), at If 13. Moreover, a warning directly to crews of the RA-5C aircraft, assuming that they are the “users” of whom section 388 speaks, would not have prevented the accidents here, since the crews had no choice but to fly aircraft equipped with the HS-1A system. Under these circumstances the pilots had no alternative to using the system in the case of a midair accident.15
B. Section 389
Section 389 provides plaintiffs-appellants no better support. Under it, a supplier of a chattel who has informed the person to whom the chattel has been supplied of its dangerous propensities may be held liable to persons who are ignorant of the dangerous character of the chattel if the supplier is aware that the chattel is unlikely to be made reasonably safe before being put to its expected use.16
One difficulty in applying section 389 to the facts of these cases is that it presupposes that use by the Navy of the chattel, as supplied by Rockwell, was improper. This we cannot say without assuming the role of naval officers. As we pointed out above, the Navy evaluated the system after the accidents at issue here and continued its use. Section 389 also presupposes that the pilots, the “persons who are ignorant of the dangerous character of the chattel,” would not fly if they knew the “dangerous character” of the “chattel.” This cannot be assumed. Naval pilots are required to fly as ordered. The alternative to ejection following an accident usually will be death.17 Finally, section 389 only applies when a chattel is not “reasonably safe.” But, to repeat, the Navy’s post-accident study found that the HS-1A system was reasonably safe. It is not for us to interfere in the Navy’s evaluation of its own weapons systems. See Aero Corp. v. Department of the Navy, 493 F.Supp. 558, 567 (D.D.C.1981).
Military personnel frequently have been sent to their deaths by the incompetence of others. Hardly a page of history lacks an example or two. We do not suggest that is the case here. However, should it be so those who serve the United States in an active military capacity are assured their survivors will receive some compensation. We merely hold that it is not for this court to increase that compensation in the manner plaintiffs-appellants suggest.
The judgment of the district court is reversed and these cases remanded for proceedings consistent with this opinion.
These holdings, to repeat, make it unnecessary for us to consider the appeals by the plaintiffs-appellants.
REVERSED and REMANDED.
. Plaintiffs-appellants contend that jurisdiction should also be based on general maritime law, see Moragne v. States Marine Lines, 398 U.S. 375, 90 S.Ct. 1772, 26 L.Ed.2d 339 (1970), which allows litigants in a wrongful death action to recover for loss of society, comfort, and care. See Sea Land Services v. Gaudet, 414 U.S. 573, 94 S.Ct. 806, 39 L.Ed.2d 9 (1974). We need not address this contention since we do not reach the damages issue in this case, and, apart from damages, the relevant law is the same under general maritime law as under the Death on the High Seas Act.
. Plaintiffs-appellants claim that the district • court erred by refusing to award them prejudgment interest, by misapplying the collateral source rule and thereby wrongly reducing the award of damages, and by declining to award compensation for the loss of services and society. Rockwell contends that the district court erred in its findings of fact, evidentiary rulings, and opinion format.
. Section 402A states:
(1) One who sells any product in a defective condition unreasonably dangerous to the user or consumer or to his property is subject to liability for physical harm thereby caused to the ultimate consumer, or to his property, if
(a) the seller is engaged in the business of selling such a product, and
(b) it is expected to and does reach the user or consumer without substantial change in the condition in which it is sold.
(2) The rule stated in Subsection (1) applies although
(a) the seller has exercised all possible care in the preparation and sale of his product, and
(b) the user or consumer has not bought the product from or entered into any contractual relation with the seller.
. Strict liability does not apply to every accident-producing activity. See, e.g., Restatement (Second) of Torts § 402A, comment k (unavoidably unsafe products); La Rosa v. Superior Court, 122 Cal.App.3d 741, 176 Cal.Rptr. 224 (1981) (used products); Pena v. Sita World Travel, Inc., 88 Cal.App.3d 642, 152 Cal.Rptr. 17 (1978) (services); Silverhart v. Mount Zion Hospital, 20 Cal.App.3d 1022, 98 Cal.Rptr. 187 (1971) (sales-service hybrid).
. Rockwell preserved its right to rely on the government contractor defense by asserting it in a motion for summary judgment.
. Most of the cases cited by plaintiffs-appellants in which a court held a manufacturer liable to a serviceman involve manufacturing, rather than design defects. See, e.g., Foster v. Day & Zimmermann, 502 F.2d 867 (5th Cir. 1974); Whitaker v. Harvell-Kilgore Corp., 418 F.2d 1010 (5th Cir.1969); Montgomery v. Goodyear Tire & Rubber Co., 231 F.Supp. 447 (S.D. N.Y.1964). In other cases relied on by plaintiffs-appellants, the parties failed entirely to raise the contractor issue. See, e.g., Stencel Aero Engineering Corp. v. United States, 431 U.S. 666, 97 S.Ct. 2054, 52 L.Ed.2d 665 (1977); Boeing Airplane Co. v. Brown, 291 F.2d 310 (9th Cir.1961); Renner v. Rockwell International Corp., 403 F.Supp. 849 (C.D.Cal.1975), vacated, 587 F.2d 1030 (9th Cir.1978).
. The recent decision of the Supreme Court in Lockheed Aircraft Corp. v. United States, - U.S. -, 103 S.Ct. 1033, 74 L.Ed.2d 911 (1983), does not detract from the Feres-Stencel doctrine’s rationales.
In Lockheed, a civilian employee of the Navy was killed in a crash of a C-5A aircraft manufactured by Lockheed, and flown by the Air Force. The United States paid death benefits to the employee’s survivors under the Federal Employees’ Compensation Act (FECA). The survivors then sued Lockheed, and Lockheed impleaded the United States as a third party defendant for indemnification purposes. The Court held that FECA does not bar such an indemnity action against the United States.
The Court distinguished the Feres-Stencel doctrine from Lockheed. The Court explained that while the United States had waived its immunity from an indemnity action based on an injury to a civilian employee who had been compensated under FECA, the United States still enjoyed sovereign immunity under Feres-Stencel from an indemnity action based on an injury to a serviceman. 103 S.Ct. at 1037-38 n. 8.
. Also, allowing liability would, as the Stencel court feared, subject the United States indirectly to paying for damages to injured servicemen, where the amount of damages would vary depending on the applicable law. 431 U.S. at 672, 97 S.Ct. at 2058.
. Plaintiffs-appellants also rely on Montgomery v. Goodyear Tire & Rubber Co., 231 F.Supp. 447 (S.D.N.Y.1964). That case, as noted above, involves a manufacturing, not a design defect.
. For convenience, we subsume the most important justifications for strict liability into four categories, based on the usage of the Oregon Supreme Court. See, e.g., Tillman v. Vance Equipment Co., 286 Or. 747, 752-54, 596 P.2d 1299, 1302-03 (1979). For a fuller list of the possible rationales for strict liability, see Henderson, Extending the Boundaries of Strict Products Liabihty: Implications of the Theory of the Second Best, 128 U.Pa.L.Rev. 1036 (1980); Owen, Rethinking the Policies of Strict Products Liability, 33 Vand.L.Rev. 681 (1980).
Some commentators include among the principal justifications for imposing strict liability the reduction of transaction costs by relieving the plaintiff of the problem of proving negligence or warranty violations. See, e.g., Schwartz, Forward: Understanding Products Liability, 67 Calif.L.Rev. 435, 459-60 (1979). Other authorities maintain that the burden of proving negligence is not significantly more difficult than that of strict liability. See, e.g., Sachs, Negligence or Strict Product Liability: Is There Really a Difference in Law or Economics?, 8 Ga.J. Int’l & Comp.L. 259 (1978).
. The Veterans’ Benefits Act may not provide full compensation for a loss in a situation similar to the present case. For example, veterans’ benefits do not compensate for loss of companionship or services. But many of these benefits are also unavailable under the Death on the High Seas Act. Moreover, compensation under the Veterans’ Benefits Act is not reduced by the high transaction costs present in ordinary products liability litigation. Stencel, 431 U.S. at 673, 97 S.Ct. at 2058. See Note, 23 B.C.L.Rev. 1025, 1083-84 (1982).
. The district court did not make a factual finding on this issue in denying Rockwell’s motion for summary judgment on the government contractor defense, R.T. at 25, or in its opinion. Plaintiffs-appellants are mistaken in their assertion to the contrary.
. Section 388. Chattel Known to be Dangerous for Intended Use.
One who supplies directly or through a third person a chattel for another to use is subject to liability to those whom the supplier should expect to use the chattel with the consent of the other or to be endangered by its probable use, for physical harm caused by the use of the chattel in the manner for which and by a person for whose use it is supplied, if the supplier (a) knows or has reason to know that the chattel is or is likely to be dangerous for the use for which it is supplied, and (b) has no reason to believe that those for whose use the chattel is supplied will realize its dangerous condition, and (c) fails to exercise reasonable care to inform them of its dangerous condition or of the facts which make it likely to be dangerous.
. It is necessary here to distinguish between the Navy, the principal user of the HS-1A system, and plaintiffs-appellants’ decedents, who were “secondary users,” or “ultimate consumers” of the system. In the case of section 388, a warning to the Navy would have been sufficient to relieve Rockwell of liability for injury to a Navy pilot. Section 388, comment n. Cf. Dalke v. Upjohn Co., 555 F.2d 245, 247-48 (9th Cir.1977).
. The district court found that both aircraft were out of control, and were flying at speeds in excess of 400 knots.
. Section 389. Chattel Unlikely to be Made Safe for Use.
One who supplies directly or through a third person a chattel for another’s use, knowing or having reason to know that the chattel is unlikely to be made reasonably safe before being put to a use which the supplier should expect it to be put, is subject to liability for physical harm caused by such use to those whom the supplier should expect to use the chattel or to be endangered by its probable use, and who are ignorant of the dangerous character of the chattel or whose knowledge thereof does not make them contributorily negligent, although the supplier has informed the other for whose use the chattel is supplied of its dangerous character.
. But see § 389, comment d (railroad is liable to soldier for accident caused by defective track even though soldier was ordered to take train over the track). The example in comment d is based on Bryson v. Hynes, 268 F. 290 (4th Cir.1920), a case which is not law in this circuit, and which is unlikely to be followed by modern courts, especially after the Feres and Stencel decisions.