POSNER, Circuit Judge.
This appeal raises difficult questions of admiralty law and sovereign immunity arising from the effort of two firms to obtain indemnity from the United States. The firms are defendants in a suit brought by the widow of a member of the armed forces who was killed while on active duty. They seek indemnity on the ground that the United States was the primary wrongdoer in the accident.
Henry Hillier, a marine safety inspector in the Coast Guard with the rank of boatswain’s mate second class, was sent to one of the Great Lakes ports to monitor the discharge of a cargo of ammonia from a barge owned by Southern Towing Company. The barge had been towed to the port by a tug owned by Memphis Towing Company. He died on the barge from inhaling ammonia fumes. Since he died what is called an “operational death” while on active duty, his widow received death benefits under 38 U.S.C., ch. 13, §§ 401 et seq., a part of the Veterans’ Benefits Act, 38 U.S.C. §§ 301 et seq.
She then sued Southern and Memphis under the admiralty jurisdiction, 28 U.S.C. § 1333, alleging that her husband’s death had been due to their negligence in, among other things, failing to inspect the barge for defects and to have proper safety equipment. Southern and Memphis impleaded the United States under Fed.R.Civ.P. 14(c) and the Suits in Admiralty Act, 46 U.S.C. §§ 741 et seq., alleging that Hillier’s death had been due not to their negligence but to the Coast Guard’s negligence in failing to train Hillier properly and provide him with adequate safety equipment. They asked that the United States be ordered to indemnify them for any damages they might be ordered to pay Mrs. Hillier. The district court granted the government’s motion for summary judgment, and dismissed the third-party complaint, on the ground that there is no right to indemnity from the government for the consequences of its negligence toward a member of the armed forces. Southern and Memphis appeal this dismissal under 28 U.S.C. § 1292(a)(3).
The government bases its argument against indemnity on two Supreme Court decisions—Feres v. United States, 340 U.S. 135, 71 S.Ct. 153, 95 L.Ed. 152 (1950), which held that the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671 et seq., does not authorize servicemen on active duty to sue the government, and Stencel Aero Engineering Corp. v. United States, 431 U.S. 666, 97 S.Ct. 2054, 52 L.Ed.2d 665 (1977), which held that a private defendant in a serviceman’s tort suit could not get indemnity from the government by showing that the government had been guilty of active wrongdoing compared to its own merely passive wrongdoing. But neither case involved an admiralty claim. Weyerhaeuser S.S. Co. v. United States, 372 U.S. 597, 83 S.Ct. 926, 10 L.Ed.2d 1 (1963), did; and the Supreme Court allowed the tort defendant to obtain contribution — partial indemnity— against the United States even though the tort plaintiff, like the military tort plaintiffs in Feres and Stencel Aero, could not have brought a tort action against the United States, instead being confined to a form of workmen’s compensation (the Federal Employees’ Compensation Act, now 5 U.S.C. §§ 8101 et seq., corresponding to the Veterans’ Benefits Act in Feres, Stencel Aero, and the present case). Although Weyerhaeuser involved a collision, which as we shall see could make a difference, it has been followed in two noncollision indemnity cases — Wallenius Bremen G.m.b.H. v. United States, 409 F.2d 994 (4th Cir.1969), which like Weyerhaeuser involved a civilian employee, and Wellington Transport. Co. v. United States, 481 F.2d 108 (6th Cir.1973), which involved a military employee and is indistinguishable from the present case. But Wellington preceded Stencel Aero, and if Stencel Aero limits the application of Weyerhaeuser to civilian cases, then Wellington can no longer be considered authoritative. The appellants also cite Ionian Glow Marine, Inc. v. United States, 670 F.2d 462 (4th Cir.1982), and Lockheed Aircraft Corp. v. United States,- U.S. -, 103 S.Ct. 1033, 74 L.Ed.2d 911 (1982). In Ionian Glow a navy ship and a private ship collided. The owner of the private ship [718]*718sued the United States to divide the damages from the collision according to the relative fault of the two ships and was allowed to include in the damages to be divided the money it had paid an injured serviceman in settlement of his tort claim. Lockheed, which differs from Stencel Aero only in involving a civilian rather than military employee of the armed forces, holds that the Federal Employees’ Compensation Act does not bar an action for indemnity against the United States.
Although the case law the parties have cited us to may seem a welter of inconsistent decisions, with the majority supporting the appellants’ claim, all except Wellington can be reconciled not only with each other but with the district court’s decision in this case; and Wellington had been rejected in other circuits even before it was (as we shall see) implicitly overruled by the Supreme Court in Stencel Aero. The key to the reconciliation is that the indemnity claim in this case is based on a tort rather than contract theory of indemnity, which is unusual in an admiralty case.
Most reported admiralty indemnity cases are cases where a shipowner, having been held liable to a longshoreman for unseaworthiness under the doctrine of Seas Shipping Co. v. Sieracki, 328 U.S. 85, 66 S.Ct. 872, 90 L.Ed. 1099 (1946), sought indemnity from the stevedore company that employed the longshoreman even though there was no explicit indemnity contract. Since the doctrine of unseaworthiness made the ship’s owner liable for any injury due to a defective or unreasonably dangerous condition on the ship and thus resembled the doctrine of strict tort liability for defective products, the duty of a negligent stevedore to indemnify the shipowner could have been based on the ground that the stevedore’s fault was “active” and the shipowner’s merely “passive.” Compare the strict-liability cases collected in Patton, Comparative Causation, Indemnity, and the Allocation of Losses Between Joint Tortfeasors in Products Liability Cases, 10 St. Mary’s L.J. 587, 598-610 (1979). In a few longshoremen cases the active-passive distinction was the ground — a tort ground — for ordering indemnity. See United States v. Rothschild Int’l Stevedoring Co., 183 F.2d 181, 182 (9th Cir.1950); States S.S. Co. v. Rothschild Int’l Stevedoring Co., 205 F.2d 253, 255 (9th Cir. 1953); Parenzan v. Iino Kaiun Kabushiki Kaisya, 251 F.2d 928, 930 (2d Cir.1958).
But in Ryan Stevedoring Co. v. Pan-Atlantic S.S. Corp., 350 U.S. 124, 133-34, 76 S.Ct. 232, 237, 100 L.Ed. 133 (1956), and cases following it such as Crumady v. “Joachim Hendrik Fisser”, 358 U.S. 423, 428, 79 S.Ct. 445, 448, 3 L.Ed.2d 413 (1959); Italia Societa per Azioni di Navigazione v. Oregon Stevedoring Co., 376 U.S. 315, 321, 84 S.Ct. 748, 752, 11 L.Ed.2d 732 (1964), and Federal Marine Terminals, Inc. v. Burnside Shipping Co., 394 U.S. 404, 420-21, 89 S.Ct. 1144, 1153, 22 L.Ed.2d 385 (1969), the Supreme Court took a different tack: it based the shipowner’s right of indemnity on the existence of an implied warranty of the stevedore to the shipowner to perform stevedoring services in a workmanlike manner. This approach has a limitation that is relevant to the present case: a warranty requires the existence of some kind of contractual relationship between indemnitor and indemnitee, although under modern notions privity of contract between the two is not necessary. See, e.g., Waterman S.S. Corp. v. Dugan & McNamara, Inc., 364 U.S. 421, 81 S.Ct. 200, 5 L.Ed.2d 169 (1960); United States v. San Francisco Elevator Co., 512 F.2d 23, 27 (9th Cir.1975).
The 1972 amendments to the Longshoremen’s and Harbor Workers’ Compensation Act substituted negligence for unseaworthiness as the standard of liability in actions by longshoremen against shipowners and abolished the shipowner’s right of indemnity against the stevedore. See 33 U.S.C. § 905(b). The abolition of indemnity was consistent with the acceptance of the active-passive distinction as the basis for indemnity; since the longshoreman now had to prove the shipowner’s negligence, it would no longer have been easy to distinguish the “passive” shipowner from the “active” stevedore. Since 1972, indemnity has continued to be sought in cases not involv[719]*719ing longshoremen and hence not within the scope of the Longshoremen’s and Harbor Workers’ Compensation Act. See, e.g., Magnum Marine v. Kenosha Auto Transport Corp., 481 F.2d 933 (5th Cir.1973) (per curiam); Todd Shipyards Corp. v. Turbine Service, Inc., 674 F.2d 401, 416-17 (5th Cir. 1982); In re Oil Spill by Amoco Cadiz, 699 F.2d 909, 915 (7th Cir.1983). But in only a few cases besides Wallenius, Wellington, and the present case has it ever been sought in the absence of a contractual relationship, direct or indirect, between the joint tortfeasors. Since Hillier was inspecting Southern’s barge pursuant to statute (the Dangerous Cargo Act, 46 U.S.C. §§ 170-170b) rather than to any contract between him (or the Coast Guard) and the appellants, the appellants cannot get indemnity — regardless of the bearing, if any, of Feres or Stencel Aero on this case — otherwise than on a pure tort theory of maritime indemnity, that is, a theory allowing indemnity between joint tortfeasors who have no contractual relationship.
There is a question whether pure tort immunity exists as a doctrine of admiralty law, and while we do not have to decide the question in this case, we do not want to leave the impression that the question is a closed one. Unlike contribution among joint tortfeasors, indemnity shifts the whole of the damage liability from one joint tortfeasor to another, just as contributory negligence shifts the whole of the burden of the accident from the injurer to the victim (and last clear chance shifts it back again). These shifts are congenial to the all-or-nothing spirit of the common law (“The common law refuses to apportion damages which arise from negligence,” Beach, Contributory Negligence 12 (2d ed. 1892)), which traditionally allowed indemnity in limited circumstances, but never contribution, see Prosser, Handbook of the Law of Torts 310-13 (4th ed. 1971), and to the notions of deterrence that animate that law. See, with reference to contributory negligence, Schofield, Davis v. Mann: Theory of Contributory Negligence, 3 Harv.L. Rev. 263, 267 (1890), and with reference to the no-contribution rule Landes & Posner, Joint and Multiple Tortfeasors: An Economic Analysis, 9 J.Legal Stud. 517, 521-26, 529 (1980). But the rule in admiralty has long been comparative rather than contributory negligence, Steamer Max Morris v. Curry, 137 U.S. 1, 14-15, 11 S.Ct. 29, 32-33, 34 L.Ed. 586 (1890), and, especially but not exclusively in collision cases, contribution rather than no contribution among joint tortfeasors. See United States v. Reliable Transfer Co., 421 U.S. 397, 401 n. 3, 95 S.Ct. 1708, 1710 n. 3, 44 L.Ed.2d 251 (1975); Cooper Stevedoring Co. v. Kopke, 417 U.S. 106, 94 S.Ct. 2174, 40 L.Ed.2d 694 (1974); Adams v. Texaco, Inc., 640 F.2d 618, 621 (5th Cir. 1981).
There is tension between contribution and indemnity. It is illustrated by Dole v. Dow Chem. Co., 30 N.Y.2d 143, 331 N.Y.S.2d 382, 282 N.E.2d 288 (1972), where the adoption of contribution in nonmaritime tort cases led New York’s highest court to reject indemnity in favor of contribution in a case where under traditional principles indemnity would have been allowed, and to quote approvingly a commentator’s description of the active-passive distinction as one of the “artificial distinctions” that had “evolved in the unnatural surroundings of an inflexible rule against contribution.” 30 N.Y.S.2d at 150, 331 N.Y.S.2d at 387, 282 N.E.2d at 293. The question whether there is ever pure tort indemnity in maritime cases was an open one in the Supreme Court even before Cooper Stevedoring Co. v. Kopke, supra, made clear that contribution is the general rule in noncollision as well as collision cases. The Court had expressly reserved the question of tort indemnity in the implied warranty cases. See Ryan, 350 U.S. at 133, 76 S.Ct. at 237; Federal Marine Terminals, 394 U.S. at 420-21, 89 S.Ct. at 1153; see also Weyerhaeuser, 372 U.S. at 603, 83 S.Ct. at 930. On the other hand, if admiralty law really were hostile to the all-or-nothing approach characteristic of indemnity, the Supreme Court probably would not have invented the stevedore’s implied warranty. That warranty, under the regime of Sieracki, allowed indemnity despite the absence of a contract [720]*720to indemnify, and did so for the same reason that the common law requires the active joint tortfeasor to indemnify the passive one: “liability should fall upon the party best able to reduce the likelihood of injury.” Italia Societa per Azioni di Navigazione, supra, 376 U.S. at 324, 84 S.Ct. at 754. Moreover, many cases before and after the era of implied warranty in longshoremen cases, and even after Kopke, have used the active-passive distinction to shift the entire liability from one maritime joint tortfeasor to another. Besides the Rothschild cases and Parenzan, cited earlier, see, e.g., Oceanic Steam Nav. Co. v. Compania Transatlántica Espanola, 144 N.Y. 663, 668, 39 N.E. 360, 371 (1895); Standard Oil Co. v. Robins Dry Dock & Repair Co., 32 F.2d 182 (2d Cir.1929); Seaboard Stevedoring Corp. v. Sagadahoc S.S. Co., 32 F.2d 886 (9th Cir.1929); Tri-State Oil Tool Indus., Inc. v. Delta Marine Drill Co., 410 F.2d 178, 181-86 (5th Cir.1969); Williams v. Brasea, Inc., 497 F.2d 67, 75-76 (5th Cir.1974); E.I. DuPont de Nemours & Co. v. Riverway Harbor Serv. St. Louis, Inc., 639 F.2d 404, 406-07 (8th Cir.1981); White v. Johns-Manville Corp., 662 F.2d 243, 247 (4th Cir.1981).
True, there is a countercurrent in the cases, illustrated by Schwartz v. Compagnie General Transatlantique, 405 F.2d 270, 272 (2d Cir.1968), where a federal immigration inspector sued the owner of a French ship, alleging “that while working in the main lounge clearing passengers for admission into the United States, he was caused to trip and fall by reason of the ship’s maintaining its piano platform and the carpeting thereof in a dangerous and defective condition.” The defendant impleaded the United States, alleging that it had “agreed and was obligated by the operation of law to conduct its immigration inspection in a reasonable and proper manner,” id., which according to the third-party complaint the United States had not done. The resemblance to the present case is apparent, and the court held that the United States had no duty to indemnify the shipowner. Id. at 275-76. Penn Tanker Co. v. United States, 409 F.2d 514, 518 (5th Cir.1969), another suit in admiralty to compel the United States to pay indemnity, hints that an underlying contractual relationship is necessary; and for similar hints see United States v. Tug Colette Malloy, 507 F.2d 1019, 1023 (5th Cir. 1975); Wedlock v. Gulf Mississippi Marine Corp., 554 F.2d 240, 242 (5th Cir.1977).
The only case we have found in which maritime indemnity has been upheld in the absence of such a relationship is a rather old district court case, Davis v. American President Lines, Ltd., 106 F.Supp. 729 (N.D.Cal.1952). Yet the paucity of cases may just reflect the fact that most joint tortfeasors have a preexisting contractual relationship, direct or indirect, with each other, although it is no longer true, as it once was, that to be deemed joint tortfeasors parties must be acting in concert. See Prosser, supra, at 314-16. Most non-maritime indemnity eases are also between parties with a preexisting contractual relationship, but not all are, and one of the principal exceptions — where a municipality has a statutory duty to keep its streets in safe condition and seeks indemnity from the person who actually caused the dangerous condition that resulted in suit against the municipality, see id. at 312 — has a family resemblance to the present case. Even in New York, where the adoption of contribution led to a narrowing of the availability of indemnity, the two principles continue to exist side by side. See Dole v. Dow Chem. Co., supra, 30 N.Y.2d at 153, 331 N.Y.S.2d at 389, 282 N.E.2d at 295; Kelly v. Diesel Constr. Div. of Carl A. Morse, Inc., 35 N.Y.2d 1, 358 N.Y.S.2d 685, 315 N.E.2d 751 (1974); Robinson v. Shapiro, 646 F.2d 734, 739-40 (2d Cir.1981); cf. Prosser, supra, at 308. When the cheapest way of avoiding a careless accident is for one party to take all the precautions, the law needs a mechanism for bringing liability to bear ultimately on that party, and indemnity supplies it. Admiralty law, like common law, is concerned with safety and efficiency. The tendency to use tort liability to spread rather than to prevent losses may be stronger in admiralty than in the common law, but it is not so overpowering as to have prevented the courts from devising imaginative theories [721]*721of indemnity in a variety of often tenuously contractual maritime settings.
But the considerations that justify a doctrine of indemnity also limit its application and make the present case an improbable one for indemnity as distinct from contribution. Unless the appellants were negligent, Mrs. Hillier will not be able to obtain damages from them. While it is possible that the government was also negligent, it is unlikely that the cheapest way to have avoided the accident would have been for just one of the joint tortfeasors, the government, to take a11 the precautions. That would imply that the best method of preventing this accident would have been for the government to load up Hillier with gas masks and training manuals, while the appellants did nothing at all to prevent the ammonia fumes from escaping. As that is unlikely, this seems a ease for contribution rather than indemnity, cf. Thibodeaux v. Texas Eastern Transmission Corp., 548 F.2d 581, 585-86 (5th Cir.1977); Wedlock v. Gulf Mississippi Marine Corp., supra, 554 F.2d at 243-44; Adams v. Texaco, Inc., supra, 640 F.2d at 620, though we cannot be sure with the case in so early a stage.
The discussion to this point should have made clear two points: that we are assuming, not deciding, that Southern and Memphis have a colorable claim to indemnity under the principles of admiralty law; and that if they do it is a pure tort claim that has no contractual basis, express or implied, but instead is premised on the idea that the United States was the active wrongdoer in the accident that killed Hillier. For the United States to be the active wrongdoer, however, it must first be a wrongdoer. Slattery v. Marra Bros., Inc., 186 F.2d 134, 139 (2d Cir.1951) (L. Hand, J.). That is, it must have breached a legal duty to Hillier. See, e.g., Palsgraf v. Long Island R.R., 248 N.Y. 339, 162 N.E. 99 (1928). But as the Supreme Court pointed out in Feres, a serviceman on active duty has never had a tort right against the government. The effect of the Federal Tort Claims Act was “not the creation of new causes of action but acceptance of liability under circumstances that would bring private liability into existence.... We know of no American law which ever has permitted a soldier to recover for negligence, against either his superior officers or the government he is serving.... We find no parallel liability before and we think no new one has been created by this Act.” 340 U.S. at 141-42, 71 S.Ct. at 156-57. The Court was speaking of soldiers suing under the Federal Tort Claims Act rather than sailors suing under the Suits in Admiralty Act. But as the appellants acknowledged at oral argument, servicemen on active duty have never been allowed to bring tort suits against the government under the Suits in Admiralty Act either. Their attempts to do so have been uniformly repelled, most recently in Charland v. United States, 615 F.2d 508 (9th Cir.1980), and Cusanelli v. Klaver, 698 F.2d 82, 85 (2d Cir.1983) — the latter a suit by a Coast Guardsman. The Suits in Admiralty Act is no more a source of rights as distinct from a waiver of the government’s defense of sovereign immunity than the Federal Tort Claims Act. The latter act provides that “the United States shall be liable ... in the same manner and to the same extent as a private individual under like circumstances ...,” 28 U.S.C. § 2674, the former that the United States may be sued in cases where, “if a private person ... were involved, a proceeding in admiralty could be maintained ...,” 46 U.S.C. § 742.
In Weyerhaeuser, if it had not been for the passage of the Federal Employees’ Compensation Act the injured seaman, a civilian, could have sued the government in tort by virtue of the Public Vessels Act, 46 U.S.C. §§ 781 et seq. The only question was whether the Federal Employees’ Compensation Act, by cutting off the seaman’s tort remedy, cut off an indemnity action as well. However that question was answered, it was possible to regard the government as having committed a tort as to which the Public Vessels Act had raised the bar of sovereign immunity and the Federal Employees’ Compensation Act had changed the remedy. But that is not possible here. [722]*722Even if there were no Veterans’ Benefits Act, corresponding to the Federal Employees’ Compensation Act in Weyerhaeuser, Hillier, a serviceman, would have had no tort cause of action against the government. Therefore the appellants have no such cause of action either, for by definition they can obtain tort indemnity only from a joint tortfeasor.
This analysis reconciles Weyerhaeuser, Wallenius, and Lockheed with Stencel Aero, and shows that Wellington was overruled by Stencel Aero, since in both cases there was no tort duty owed by the government to the plaintiff and therefore no basis, in the absence of an express or implied contractual obligation, for requiring the government to indemnify the defendant. The Court in Stencel Aero noted that there was no contractual relationship between the defendant and the government and that the defendant’s theory of indemnity was that it had been passively and the United States actively negligent. See 431 U.S. at 667 n. 2, 668, 97 S.Ct. at 2056 n. 2, 2056. Although the Supreme Court did not mention Wellington, the court of appeals decision affirmed (under a different name) in Stencel Aero had discussed Wellington at length and rejected its approach, Donham v. United States, 536 F.2d 765, 773-74 (8th Cir. 1976), as had the Fifth Circuit in Certain Underwriters at Lloyd’s v. United States, 511 F.2d 159, 163 (5th Cir.1975).
Ionian Glow also is distinguishable from the present case. There was no doubt there that the United States was a joint tortfeasor. A navy vessel collided with a privately owned vessel. Both ships sustained damage; both were at fault. In such a case the total damages (including any suffered by a third party) are divided between the two ships according to their relative fault. United States v. Reliable Transfer Co., supra. Part of the damage bill happened to come from the liability of the private ship to members of the navy ship’s crew. As the navy ship was admittedly at fault, there was no problem with treating the navy as a joint tortfeasor and dividing damages in the usual fashion.
So there is no basis in conventional admiralty principles for requiring the government to indemnify the appellants in this case. But it can be argued that we should exercise our judicial creativity and, recognizing that admiralty is an evolving body of judge-made law, create a doctrine of indemnity that is neither tort-based nor contract-based but that shifts liability to the government where it would promote safety or efficiency to do so. But the fact that we are dealing with an issue of sovereign immunity makes such an exercise in creativity inappropriate. The appellants’ third-party claim can be maintained only if the United States has waived its sovereign immunity from such claims in the Suits in Admiralty Act. That act originally waived sovereign immunity only with respect to damage caused by government vessels and cargoes, but it was amended in 1960 to add the language we quoted earlier, and this and other courts have read the 1960 amendment to waive the government’s sovereign immunity with respect to any claims that could be maintained in admiralty if brought against a private person. See, e.g., Bearce v. United States, 614 F.2d 556, 558 (7th Cir.1980); De Bardeleben Marine Corp. v. United States, 451 F.2d 140, 142-45 (5th Cir.1971); National Union Fire Ins. Co. v. United States, 436 F.Supp. 1078, 1080 (M.D. Tenn.1977). A contractual claim for indemnity would be such a claim and so would a tort indemnity claim if but only if tort indemnity is a maritime as well as nonmaritime doctrine. But an indemnity claim sounding neither in tort nor in contract, such as is asserted in the present case, would not be. There is no evidence that by amending the Suits in Admiralty Act in 1960 Congress authorized the courts to subject the United States to novel and far-reaching judge-made liabilities for the conduct of its shore-based personnel. The lack of such evidence is decisive; to interpret the Suits in Admiralty Act as a waiver of the government’s sovereign immunity not in a conventional admiralty suit, but in a suit based on a radically new theory of liability such as the appellants want us to invent to decide this case, would violate the [723]*723principle that the United States’ consent to be sued is “not lightly inferred.. .. ” United States v. Mitchell, - U.S. -, -, 103 S.Ct. 2961, 2968, 77 L.Ed.2d 580 (1983). See also United States v. Sherwood, 312 U.S. 584, 590, 61 S.Ct. 767, 771, 85 L.Ed. 1058 (1941); United States v. Erika, Inc., 456 U.S. 201, 102 S.Ct. 1650, 72 L.Ed.2d 12 (1982).
Yet if Hillier had been a civilian employed by the Defense Department, Lockheed would allow the appellants to seek indemnity from the United States. Although Hillier was technically a member of the armed forces — and more than technicalities are involved, since his widow did receive compensation under the Veterans’ Benefits Act, as only a serviceman’s widow could — the work he did, inspecting civilian cargoes discharged in Great Lakes ports, does not have a martial air about it. It could just as well have been performed by a civilian as by a serviceman. Why should it matter whether he is the one or the other?
The Court in Feres and Steneel Aero did point out that since the Federal Tort Claims Act adopts the law of the state where the tort occurred, 28 U.S.C. § 1346(b), the government’s liability to third parties in connection with accidents to soldiers and airmen would differ depending on where the accident occurred. But admiralty law is a uniform body of federal law, at least in theory, though in practice there are variations among the federal circuits. And while the nonuniformity argument of Feres and Steneel Aero can be criticized — if the government can cope with the laws of the different states with regard to its liability under the Tort Claims Act for indemnity in connection with accidents to any of its 2 million civilian employees (Lockheed), it should be able to cope with similar suits arising from accidents to any of its 2 million servicemen — it was accepted by the Supreme Court in Steneel Aero and is unavailable to the government here. Another point made in Steneel Aero, that indemnity would allow an end run around the limitations on government liability in the Veterans’ Benefits Act, is undermined by the rejection in Lockheed of a similar argument under the Federal Employees’ Compensation Act.
A concern that remains significant here, however, is the possible adverse impact on military efficiency if the government’s negligence to a serviceman were a litigable issue — albeit not in a tort suit by the serviceman against the government, because he cannot bring such a suit. The principle of automatic, even unquestioning, obedience to the commands of superior officers remains fundamental to the operation of the military services, in this country as in the rest of the world, in peacetime as in wartime, in the support as in the combat branches. See Chappell v. Wallace, - U.S. - , -, 103 S.Ct. 2362, 2365-67, 76 L.Ed.2d 586 (1983). The principle is compromised if a soldier or sailor can be required to testify in court against his superiors, which no doubt is what would happen in this case if the third-party claims went to trial. That it could also happen in suits under the Suits in Admiralty or Public Vessels Acts brought by private victims of the Coast Guard’s negligence shows only that this principle is not absolute. Even though the Air Force can be sued under the Tort Claims Act by civilians whom it injures in accidents, the Supreme Court in Steneel Aero was concerned with the incremental damage to the obedience principle of allowing indemnity suits based on the Air Force’s negligence. Speculative as this concern may seem to some, the Supreme Court has reiterated it in a variety of military settings over many years, see, e.g., In re Grimley, 137 U.S. 147, 153, 11 S.Ct. 54, 55, 34 L.Ed. 636 (1890); Rostker v. Goldberg, 453 U.S. 57, 65-66, 101 S.Ct. 2646, 2651-52, 69 L.Ed.2d 478 (1981) — most recently in Chappell v. Wallace, supra, decided just the other day — and we may not ignore it.
The Coast Guard is a military service, 14 U.S.C. § 1, and long has been, see Evans, The United States Coast Guard 1790-1915: A Definitive History (1949); and its organization along military lines, while maybe not inevitable, is not without purpose. Among other things the Coast [724]*724Guard becomes a part of the Navy in wartime, 14 U.S.C. § 3 — it saw action for example in the early days of the war in Vietnam. See Bloomfield, The Compact History of the United States Coast Guard 286 (1966). True, its function even in wartime is mainly support rather than combat, and Mr. Hillier was not in a combat specialty; but in a modern army and navy, most of the soldiers and sailors are support rather than combat troops, yet are, as we have pointed out, subject to the same principle of automatic obedience as combat troops. In any event, there should be only one judge-made rule for the military; it is not within the judicial competence to decide exactly where in the military services the principle of obedience should be preserved.
Even if we thought the distinction between military (this case) and civilian (Lockheed) employees of the armed forces a tenuous one, this would not change our decision. To the argument in Lockheed that Stencel Aero precluded indemnity, the Supreme Court replied: “The issue in Stencel, again relating to the underlying substantive claim, was whether the Government's waiver of sovereign immunity in the Federal Tort Claims Act applied to an indemnity action based on an injury to a serviceman. Relying primarily on the military nature of the action, we held that the doctrine of Feres v. United States, 340 U.S. 135, 71 S.Ct. 153, 95 L.Ed. 152 (1950), precluded the substantive claim without regard to any exclusive liability provision.” 103 S.Ct. at 1037-38 n. 8. The two sentences in this passage represent two different ways of stating the distinction between the present case and the cases (all but Wellington ) on which the appellants rely. First, as the government has never had a tort duty to its sailors, an indemnity action based on the theory that the government was the primary or active tortfeasor in an accident to a sailor has no basis in admiralty law and there is therefore no tort right the bar to enforcement of which the Suits in Admiralty Act could have lifted. Second, by interpreting Stencel Aero as having been based “primarily on the military nature of the action,” footnote 8 of Lockheed suggests that Stencel Aero rests on the absence of a legal wrong by the government rather than on practical considerations that might argue for limiting a liability that would otherwise exist in law. Unless and until Stencel Aero is overruled — and not only are we not alone in taking Lockheed at face value as distinguishing rather than overruling Stencel Aero, see McKay v. Rockwell Int’l Corp., 704 F.2d 444, 448 and n. 7 (9th Cir.1983), but the Supreme Court cited Stencel Aero with approval just the other day, in the Chappell case,-U.S. at-, 103 S.Ct. at 2364— to allow maritime tort defendants such as the appellants in this case to get indemnity from the United States would erase one artificial distinction only to create another: between accidents to servicemen on land and at sea.
The judgment dismissing the third-party complaints against the United States is
Affirmed.