Twenty-six years ago, in Ira S. Bushey & Sons, Inc. v. United States, 398 F.2d 167 (2d Cir.1968), this court held that the United States Government was vicariously liable for damage to a drydock caused by a drunken sailor who was returning to ship from a night's liberty. In his celebrated opinion, Judge Henry Friendly described the basis of respondeat superior as the "deeply rooted sentiment that a business enterprise cannot justly disclaim responsibility for accidents which may fairly be said to be characteristic of its activities." Id. at 171. Even though the sailor had become drunk while on liberty and far off base, we noted that drinking on leave was so common a part of naval life that the sailor's drunken return to ship could fairly be deemed to be characteristic of the military enterprise and, hence, that the government should be held liable for the damage that he caused. See id. at 172.

In Bushey, we applied admiralty law. Today--in a case that again involves a seaman who had too much to drink--we must apply the law of Guam. This, in turn, points us to California decisions for guidance. As it happens, California had taken the lead in developing the modern law of respondeat superior even before Bushey. And, so, rounding out the circle, we now reach the same conclusion as did Judge Friendly, twenty-six years ago.

Although in Bushey the decision that the government was liable under respondeat superior pretty well disposed of all the issues in the case, a similar conclusion here offers no such closure. Instead, it forces us into that singular tangle of seemingly inconsistent rulings and rationales known as the Feres doctrine. In the end, we conclude that the most sensible reading of Feres and its progeny does not bar this suit. But we would be less than candid if we did not admit that the Feres doctrine has gone off in so many different directions that it is difficult to know precisely what the doctrine means today.

The facts are simple enough and not disputed. On the morning of April 13, 1985, Robert S. Maine, ("Maine") a Navy serviceman on active duty at the U.S. Naval Ship Repair Facility on the island of Guam, went on liberty after having completed a grueling 24 hour duty shift. While on liberty he was free to leave the base as he pleased and travel up to 50 miles away. He could also be recalled for duty at any time.

Maine decided to have a good time. By noon, he was relaxing at an on-base beach party and drinking beer with Navy friends. Later that afternoon, he purchased two six-packs of beer at the base PX with his Navy comrade, Karin Conville ("Conville"), and returned with her to his barracks to drink several more cans. At dinnertime, Maine accompanied friends to the enlisted men's club, where he consumed two cocktails with his meal. After dinner, he attended a barracks party in the room of a superior officer, with several other superior officers present. There, Maine drank three or four more beers and--when he left to return to his own barracks at about 11:00 p.m.--Conville and another Navy comrade named Jean Buquet noticed that he seemed to be drunk. At around 11:30 p.m., Maine had difficulty sleeping and decided to drive off base to get something to eat. Feeling tired, he aborted his snack mission and tried to return to base. On the way back, he caused the accident that injured Scott A. Taber ("Taber").

Taber was an enlisted Seabee--a construction worker in the United States Navy--and was stationed at Camp Covington, Guam. At 6:00 p.m. on Friday, April 12th, he too went on liberty. Accordingly, he was free to go off base at any time, to travel anywhere within 50 miles of his base and, unless he was recalled for duty, to do as he pleased until his liberty ended at 6:00 a.m. on the following Monday.

Around 2:00 p.m. on Saturday April 13, Taber's civilian friend, Estelita Stills ("Stills"), met Taber at his base in her car. They planned to spend the weekend together at her house, which was located off the base. Before going there, however, the two drove to her cousins' home for dinner at the nearby U.S. Naval Station. There, Taber enjoyed a meal and, as a friendly gesture in return, helped fix the cousins' car. Shortly before midnight, Stills and Taber left for Stills's house and their weekend of rest and recreation. As fate would have it, they never got there. While they were driving on the public roadway toward Stills's house, Maine crashed into them, injuring Taber severely.

Two years later, Taber started this action for damages under the Federal Tort Claims Act ("FTCA"), 28 U.S.C. Secs. 1346(b), 2671, in the United States District Court for the Western District of New York, (David G. Larimer, Judge ). Naming both Maine and the United States Government as defendants, Taber complained that he was injured as a result of Maine's negligent driving and that, because Maine was acting within the scope of his Naval employment when he caused the accident, the government was liable on a theory of respondeat superior. The government moved for summary judgment on the grounds that, as a matter of law, Maine's conduct fell outside the scope of his military service and that, therefore, the government was not liable for Taber's injuries.

Taber opposed the government's motion and cross-moved to amend his complaint. The proposed amended complaint claimed that the government was vicariously liable for the actions of the Navy personnel who allegedly had negligently allowed Maine to get drunk and to drive off base. In response, the government argued that the doctrine established by Feres v. United States, 340 U.S. 135, 71 S. Ct. 153, 95 L. Ed. 152 (1950), which held that "the Government is not liable under the Federal Tort Claims Act for injuries to servicemen where the injuries arise out of or in the course of activity incident to service," id. at 146, 71 S.Ct. at 159, barred Taber's new claims.

The district court agreed with the government in all respects. In his Decision and Order dated December 7, 1993, Judge Larimer granted summary judgment to the government because "Maine's drunk driving incident on April 13, 1985, was not in the line of duty and therefore the United States is not liable under the doctrine of respondeat superior." He also denied Taber's motion to amend his complaint holding that these claims would be barred by the Feres doctrine.

The action proceeded against Maine, however. After a bench trial in which Maine appeared pro se, the district court found Maine liable for negligence and assessed Taber's damages at $300,000. A final judgment was entered and Taber appealed.

In his appeal, Taber presses only his original claim that the government is vicariously liable for Maine's negligence, and abandons his motion to amend his complaint. He argues that: (1) the district court erred in failing to apply the doctrine of respondeat superior to Maine's drunk driving; and (2) the Feres doctrine does not bar this claim. We agree with both of Taber's contentions. Accordingly, we reverse the district court's judgment and remand the case for further proceedings.

The FTCA allows civil actions against the government based on the negligent acts or omissions of its employees, see 28 U.S.C. Sec. 1346(b), including those of members of the Armed Services who are acting "in the line of duty." 28 U.S.C. Sec. 2671. The courts have uniformly equated the FTCA's "line of duty" language with the phrase "scope of employment," as that concept is defined by the respondeat superior law of the jurisdiction in which the accident occurred. See McHugh v. University of Vermont, 966 F.2d 67, 75 n. 9 (2d Cir.1992) (citing McCall v. United States, 338 F.2d 589 (9th Cir.1964), cert. denied, 380 U.S. 974, 85 S. Ct. 1334, 14 L. Ed. 2d 269 (1965)); Merritt v. United States, 332 F.2d 397, 398 (1st Cir.1964). Because the accident in this case happened in Guam, we must follow Guam's law of respondeat superior. Since the law of Guam is anything but certain, however, that is easier said than done.

Where the law of Guam is unclear, the Ninth Circuit, serving as Guam's highest appellate court, see 48 U.S.C. Sec. 1424-2 (investing the Ninth circuit with certiorari jurisdiction to review "all final decisions of the highest court of Guam"), has instructed courts to look to California law for guidance. Cf. People of the Territory of Guam v. Muna, 999 F.2d 397, 399 n. 1 (9th Cir.1993) ("As a general rule, we refer to California law in resolving unsettled questions of Guam law"). This is particularly true in cases, like this one, where the relevant Guam statute is identical to, and indeed derives from, the California Civil Code. See Concepcion v. United States, 374 F. Supp. 1391, 1395 (D.Guam 1974) (noting that the respondeat superior section of the Civil Code of Guam was "taken verbatim" from the California Civil Code and that "[t]he California cases construing the doctrine ... are persuasive in the construction of similar Guam laws"); cf. People of the Territory of Guam v. Iglesias, 839 F.2d 628, 629 (9th Cir.1988) ("We should [be guided by California law] when California law forms the antecedent of a Guam statute and the intent of the Guam legislature is difficult to discern"). Indeed, on several occasions, the Ninth Circuit relied upon California cases in intuiting Guam's law of respondeat superior. See Williams v. United States, 248 F.2d 492, 494-95 (9th Cir.1957), cert. denied, 355 U.S. 953, 78 S. Ct. 537, 2 L. Ed. 2d 529 (1958); United States v. Johnson, 181 F.2d 577, 580-81 (9th Cir.1950) ("California decisions with respect to the tests for determining when the servant of a private employer is acting within the course of his employment supply a convenient starting point for a discussion").

Nevertheless, we begin with Guam. We have found only one decision by a court sitting in Guam, which addresses Guam's law of respondeat superior. See Concepcion, 374 F. Supp. at 1395.1 That case held that an off-duty Navy petty officer was not acting in the line of duty when he drove a servicemember to a naval communications station so that the servicemember could make an emergency phone call to his family. Id. at 1393. If Concepcion properly defines Guam's law of respondeat superior, we would agree with the district court that Taber's claim must fail, for the links between the tortfeasor's actions and the government are closer in that case than they are in the case before us. Consequently, the first question that we must ask is whether Concepcion is controlling today. We think it is not.

The government argues that we should follow Concepcion because we should pay particular attention to the holdings of a federal district judge regarding the law of the jurisdiction in which that federal district court sits. This deference is certainly justified when a district judge--familiar with local law and customs--bases his or her decision on that familiarity. See Stevens v. Barnard, 512 F.2d 876, 880 (10th Cir.1975) ("The views of a federal judge, who is resident of the state where the controversy arose in a case involving interpretations of the law of the state, carry extraordinary persuasive force on appeal where there are no state decisions on point or none which provide clear precedent.").

It is much less appropriate, however, when the district court does not rely on its "home grown" expertise, but, as in Concepcion, has simply interpreted a statute that, itself, originally comes from another jurisdiction, and has cited only cases from that other jurisdiction in support of its interpretation. Moreover, reliance on the local district court is least likely to be sound when, again as here, there exists a Court of Appeals (the Ninth Circuit), that hears both diversity cases from the state of the statute's origination (California) and appeals from the local and federal courts of the jurisdiction in which the district court sits (Guam). This is especially so when that appellate court has spoken frequently on the issue.

For these reasons we conclude that Concepcion, a rather dated holding of an isolated district court in Guam, which cited a few lower court California cases and no Guam cases in interpreting a statute that is identical to California's, is a weak guide to Guam law today. A better source of law would seem to be the many California and Ninth Circuit decisions that have given meaning to California's respondeat superior statute, See, e.g., Mary M. v. City of Los Angeles, 54 Cal. 3d 202, 208-09, 285 Cal. Rptr. 99, 101-02, 814 P.2d 1341, 1343-44 (1991); Perez v. Van Groningen & Sons, Inc., 41 Cal. 3d 962, 967-68, 227 Cal. Rptr. 106, 108-09, 719 P.2d 676, 677-79 (1986); Hinman v. Westinghouse Elec., Inc., 2 Cal. 3d 956, 959-60, 88 Cal. Rptr. 188, 190, 471 P.2d 988, 990 (1970); Liu v. Republic of China, 892 F.2d 1419, 1427-28 (9th Cir.1989), cert. dismissed, 497 U.S. 1058, 111 S. Ct. 27, 111 L. Ed. 2d 840 (1990); United States v. Romitti, 363 F.2d 662, 665 (9th Cir.1966), and, by derivation, the Guam statute as well. Roberto v. Aguon, 519 F.2d 754, 755 (9th Cir.1975) ("Appellate courts have consistently recognized that decisions of California courts ... subsequent to the adoption of the Guam codes while not binding, are persuasive.").

It seems clear to us that California law (and by implication the law of Guam) would hold the government vicariously liable for Maine's actions. California was one of the first states in the nation to adopt an expansive reading of the respondeat superior doctrine. As early as 1961, commentators noted that California had taken the lead in equating the scope of respondeat superior liability to the traditionally broader coverage mandated by workers' compensation statutes. Thus, California employers were subject to liability for injuries to third parties caused by the behavior of their employees whenever the employees' acts "arose out of or in the course of" their employment relationship. See Guido Calabresi, Some Thoughts on Risk Distribution and the Law of Torts, 70 Yale L.J. 499, 545 (1961).

This approach to respondeat superior is even more evident in numerous California cases decided after Concepcion (which inevitably cast doubt on Concepcion itself). For example, in Rodgers v. Kemper Construction Co., 50 Cal. App. 3d 608, 124 Cal. Rptr. 143 (4th Dist.1975), a subcontractor was held vicariously liable for an assault committed by two of its employees who had lounged around drinking for several hours in what was, ironically, called the "dry house" (a rest area/locker room located on the job site). On a Friday night after their work shift had ended, the employees, though free to go home, stayed in the dry house and got drunk. Later they went outside and got in a fight with the plaintiffs. See Id. at 615, 124 Cal.Rptr. at 146-47.

In finding respondeat superior liability, the court stated that "the inquiry should be whether the risk was one 'that may be fairly regarded as typical of or broadly incidental' to the enterprise undertaken by the employer." Id. at 619, 124 Cal.Rptr. at 149 (citations omitted). The court further noted that under California law,

where social or recreational pursuits on the employer's premises after hours are endorsed by the express or implied permission of the employer and are 'conceivably' of some benefit to the employer or, even in the absence of proof of benefit, if such activities have become 'a customary incident of the employment relationship,' an employee engaged in such pursuits after hours is still acting within the scope of his employment.

In Rodgers, the subcontractor "customarily permitted employees to remain on the premises in or about the dry house long after their work shift had ended" and it was also "customary, particularly on Friday evenings, for employees to sit around the dry house after their work shift and talk and drink beer, often ... joined by their supervisors." Id. at 619-20, 124 Cal.Rptr. at 149. Because it "was neither unusual nor unreasonable" for the assailants to be on the job site drinking before the assault, and because such drinking in the dry house "was a customary incident of the employment relationship," the court ruled that their related tortious actions fell within the scope of their employment. Id. Not surprisingly, the court in Rodgers relied heavily on our decision in Bushey. See id. at 618, 50 Cal. App. 3d 608, 124 Cal.Rptr. at 148-49.

Similarly, in Childers v. Shasta Livestock Auction Yard, Inc., 190 Cal. App. 3d 792, 235 Cal. Rptr. 641 (3d Dist.1987), Shasta's foreman gave Childers and Abbott (both Shasta employees) the keys to his office at the end of the day and told them to go have a beer. The two employees were later joined by a customer, and the three of them drank both beer and hard liquor for several hours, getting quite drunk. Id. at 799, 235 Cal.Rptr. at 642-43. At around 10:00 p.m., Abbott suggested to Childers that they drive off to feed Abbott's horses. Abbott drove her truck off the road, killing herself and injuring Childers. Id.

In addressing Childers's claim against Shasta, the court made clear that the fact that Childers's injuries occurred away from the work site did not bar the employer's vicarious liability for Abbott's drunk driving. The court said:

respondeat superior liability is properly applied where an employee undertakes activities within his or her scope of employment that cause the employee to become an instrumentality of danger to others even where the danger may manifest itself at times and locations remote from the ordinary workplace.

Id. at 804-05, 235 Cal.Rptr. at 647. The court also explicitly linked the scope of liability under respondeat superior to that which would make an employer liable to an employee under the workers' compensation laws. Id. at 801, 235 Cal.Rptr. at 644.

Consistently with these cases, the Ninth Circuit has, itself, read California's law of respondeat superior broadly. See Liu, 892 F.2d at 1427-28 (citing both Rodgers and Childers ); Romitti, 363 F.2d at 665-66 (adopting "scope of enterprise analysis" as respondeat superior standard under California law). In fact, the Ninth Circuit took specific note of both Rodgers and Childers when, in a case very close to ours, it stated in dictum that California law would likely impose liability on the government for the actions of an off-duty servicemember who becomes drunk while on a military base, and then drives off base and injures someone. See Doggett v. United States, 875 F.2d 684, 687 (9th Cir.1989).

The district court below tried to distinguish these authorities on the ground that the drinking in Rodgers and Childers took place at the work site while Maine's supposedly did not. We disagree. The drinking in both Rodgers and Childers occurred at work-site rest areas (the "dry house" and the business office, respectively)--not on the assembly line. Similarly, although Maine did not drink while working at the Naval Ship Repair Facility, he drank at an on-base beach party, at the enlisted men's club, and in the barracks--all of which were located on his base. These places were as much on-site rest areas as the ones involved in both Rodgers and Childers.3

The government understandably seeks to rely on an older conception of respondeat superior. This view of the doctrine required a close link between the acts of the "agent" and "profit" accruing to the master before vicarious liability attaches to the latter. See Restatement (Second) of Agency Sec. 228 (1984). But today this position is in hasty retreat, if not rout. Thus Rodgers and Childers held that the employer-benefit requirement is met whenever broad potential effects on morale and customer relations exist, or where the employer has implicitly permitted or endorsed the recreational practices that led to the harm. See Rodgers, 50 Cal. App. 3d at 618-21, 124 Cal.Rptr. at 149-50; Childers, 190 Cal. App. 3d at 805-06, 235 Cal.Rptr. at 647-48. The decline of this profit requirement, in a direct sense, can also be seen in the wholesale abandonment of the charitable immunity exception to respondeat superior. See W. Page Keeton, Dan B. Dobbs, Robert E. Keeton & David G. Owen, Prosser and Keeton on Torts, Sec. 133 (5th ed. 1984) (hereinafter, "Prosser & Keeton"); Restatement (Second) of Torts Sec. 895E (1977).

Of course drinking by servicemembers can be viewed as important to military morale, just as drinking was apparently instrumental to good employee morale and customer relations in Rodgers and Childers. Hence, "employer-benefit" can be adduced in all these cases. But in the end, "employer-benefit" is significant only because it is one way of showing that the harm that drinking causes can properly be considered a cost of the employer's enterprise.

California courts have said that the doctrine of respondeat superior is "concerned with the allocation of the cost of industrial injury." Childers, 190 Cal.App.3d at 801, 235 Cal. Rptr. at 644 (emphasis added). The issue is simply whether the employee's "conduct is not so unusual or startling that it would seem unfair to include the loss resulting from it among other costs of the employer's business." Rodgers, 50 Cal.App.3d at 619, 124 Cal. Rptr. at 149 (emphasis added). Thus, our focus must be on the relationship between the servicemember's behavior and the costs of the military enterprise.

Here, it is undisputed that drinking on base during off-duty hours was a commonplace, if not an officially condoned activity. It certainly was a customary incident of Maine's employment relationship with the Navy, as that element is described in Rodgers. See 50 Cal.App.3d at 620, 124 Cal. Rptr. at 150. And in the context of the military mission, an occasional drunken servicemember who leaves government premises and causes damage is a completely foreseeable event, in the sense that it is a reasonably obvious risk of the general enterprise. As such, we do not think that it would be either "unfair" or the slightest bit unreasonable to impose that cost on the government. To the contrary, given the pervasive control that the military exercises over its personnel while they are on a base, it is totally in keeping with the doctrine of respondeat superior to allocate the costs of base operations to the government. See William M. Landes & Richard A. Posner, The Positive Economic Theory of Tort Law, 15 Ga.L.Rev. 851, 914-15 (1981) (discussing respondeat superior as an incentive for employers to exert their control over employees to induce careful conduct). And this is so quite apart from whether or not the military benefits from the boost in morale achieved through fairly lenient on-base drinking policies.

As the leading Torts treatise has put it, "the integrating principle" of respondeat superior is "that the employer should be liable for those faults that may be fairly regarded as risks of his business, whether they are committed in furthering it or not." Fowler V. Harper, Fleming James, Jr. & Oscar S. Gray, The Law of Torts Sec. 26.8 (2d ed., 1986) [hereinafter, "Harper & James"]. Judge Friendly made the same point most elegantly in Bushey. "The proclivity of seamen to find solicitude by copious resort to the bottle," he wrote, "has been noted in opinions too numerous to warrant citation. Once all this is granted, it is immaterial that [the coastguardsman's] precise action was not to be foreseen." 398 F.2d at 172. After all, the government "cannot justly disclaim responsibility for accidents which may fairly be said to be characteristic of its activities." Id. at 171.

We believe the law of Guam reaches the same conclusion. Accordingly, we hold that the government is vicariously liable for Maine's conduct.

Unfortunately, this does not end our analysis. Although the government may be responsible for Maine's actions under respondeat superior, it may not be liable to Taber because of the Feres doctrine. The district court did not reach this issue since it held that the government was not vicariously liable in the first place. It did, however, conduct a Feres analysis of the proposed claims contained in Taber's cross-motion to amend his complaint. The court concluded that these claims, which sought to impose liability upon the government for the actions of Navy personnel other than Maine, were barred by the Feres doctrine.

It is not clear whether the district judge would have also ruled that the Feres doctrine barred Taber's original claim based on Maine's conduct. In denying Taber's cross-motion to amend his complaint on Feres grounds, the district court emphasized the nature of the relationship between the allegedly negligent Navy personnel (i.e., the injurers ) and the government. The district court did not focus on the nature of the relationship between Taber (i.e., the plaintiff ) and the government at the time of his injury--a consideration that is, in almost all cases, of much greater relevance to the Feres analysis.

A fair reading of the district judge's opinion suggests, however, that he would have concluded that Feres also barred Taber's current suit, although we cannot be sure. In any event, rather than remanding the case to the district court for further consideration on this point, we choose to address the Feres question directly. We do this because both parties have briefed and argued Feres at length, and because we believe we should try to clarify what has become an extremely confused and confusing area of law.4

The Feres doctrine started lucidly enough as a rule that barred servicemember's claims under the FTCA for injuries that "arise out of or are [sustained] in the course of activity incident to service." Feres, 340 U.S. at 146, 71 S. Ct. at 159. This language, which derived from the words characteristically found in both state and federal workers' compensation statutes, was not chosen accidentally. Indeed, at its inception, the rule in Feres is best understood as an attempt to preclude suits by servicemembers against the government because, as military employees, they received government disability and death benefits--benefits that the Court observed were similar to (and if anything more generous than) most civilian workers' compensation awards. See Feres, 340 U.S. at 145, 71 S.Ct. at 158-59.

Then, as now, civilian workers' compensation statutes typically barred tort suits by employees against their employers for injuries arising out of or in the course of employment. See Harper & James Sec. 11.2 ("The compensation under these [state and federal] acts is usually the exclusive remedy of the employee and dependents against the employer, in lieu of any amounts that might otherwise have been recovered in a lawsuit for injuries covered by the acts."); Arthur Larson & Lex K. Larson, The Law of Workermen's Compensation, Sec. 65.30 (1994) [hereinafter "Larson on Workmen's Compensation"]. It must have seemed reasonable to the Supreme Court to treat military employees in a similar manner. After all, treating like cases alike is the great engine of the law.

That such a reading of the FTCA was exceedingly willful, and flew directly in the face of a relatively recent statute's language and legislative history, see generally, United States v. Johnson, 481 U.S. 681, 692-703, 107 S. Ct. 2063, 2069-75, 95 L. Ed. 2d 648 (1987) (Scalia, J., dissenting), apparently did not trouble the Court much--intent, as it was, to make the FTCA "fit" the legal landscape of the time.5 Indeed, in this regard, it is particularly noteworthy that in 1949, one year before the Court decided Feres, Congress had amended the Federal Employees' Compensation Act specifically to preclude covered federal employees from maintaining personal injury and wrongful death suits under the FTCA. See Larson on Workermen's Compensation at Sec. 65.33 (citing 5 U.S.C. Sec. 8116(c)).

It may occasionally be desirable for courts to pressure legislatures to reconsider outdated statutes so that, unless the legislatures make clear their continued preference for disparate treatment, like cases may be treated alike. See generally, Guido Calabresi, A Common Law for the Age of Statutes (1982). Although apparently this was precisely what the Court was doing in Feres, its willingness to ignore language, history, and the process of incremental law making (not to mention possible ways of dialoguing with Congress to discern the legislature's actual intent) was nevertheless remarkable. In any event, none of these considerations seemed to matter to the Court which seemingly concluded that the federal "systems of simple, certain and uniform compensation for injuries or death in the armed services," Feres, 340 U.S. at 144, 71 S.Ct. at 158, should be, like workers' compensation laws, an injured servicemember's sole source of recovery. Id. at 143, 71 S.Ct. at 157-58.

Despite its willful and arguably misguided origins, Feres would have been both easy enough to understand and to follow had it actually been applied to all servicemembers who benefited from the "simple, certain, and uniform [system of government] compensation." Later courts might have taken comfort in Congress' apparent acquiescence in the Supreme Court's construction of the FTCA, see Johnson, 481 U.S. at 686 & n. 6, 107 S. Ct. at 2066 & n. 6, even though such reliance on legislative inaction is rarely sound given the degree of inertia that is intentionally built into our system of checks and balances. See id. at 702-03, 107 S.Ct. at 2075 (Scalia, J., dissenting). And in time, the willfulness of the Court's decision in Feres would have been forgotten, especially if it had achieved treatment for military personnel that was, in fact, "like" the treatment accorded to the great mass of civilian employees. But such a result seemingly did not come about.

Instead Feres quickly lurched toward incoherence. Part of the problem lay with Brooks v. United States, 337 U.S. 49, 69 S. Ct. 918, 93 L. Ed. 1200 (1949), a case that had immediately preceded Feres. In Brooks, the Supreme Court permitted recovery under the FTCA to two servicemen, the Brooks brothers, who had been "on furlough, driving along the highway, under compulsion of no orders or duty and on no military mission [when] a government owned and operated vehicle collided with [them]." Feres, 340 U.S. at 146, 71 S. Ct. at 159. Attempting to distinguish its previous holding in Brooks, the Feres court noted that "[t]he injury to Brooks did not arise out of or in the course of military duty," and that the "Brooks's relationship [to the government] while on leave was not analogous to that of a soldier injured while performing duties under orders." Id.

But the plaintiffs in Brooks were eligible for precisely the same set of government benefits as were the plaintiffs in Feres, and indeed they originally collected them in addition to receiving their FTCA awards. See Brooks, 337 U.S. at 53-54, 69 S.Ct. at 920-21 (remanding the case for further proceedings to determine whether and by how much Brooks's FTCA judgment should be off-set by their/his military disability benefits). This fact immediately placed Brooks in tension with Feres. For if the existence of a military death and disability compensation scheme was to be the singular rationale for precluding servicemembers from asserting tort claims against the government, Feres should have overruled Brooks.

But Feres neither overruled Brooks, nor limited Brooks to its immediate facts. Indeed, the Supreme Court and several circuit courts (without reproof from the Supreme Court), have subsequently applied Brooks rather than Feres, and allowed FTCA claims in a significant number of cases in which the injured plaintiffs were fully covered by the government's compensation scheme. See e.g. United States v. Brown, 348 U.S. 110, 112, 75 S. Ct. 141, 143, 99 L. Ed. 139 (1954) (Brooks held controlling where veteran sued Veteran's Administration hospital for malpractice in treating an injury sustained while plaintiff was in military service); Harvey v. United States, 884 F.2d 857, 861 (5th Cir.1989) (despite the fact that serviceman "[i]ndisputably ... received some compensation from the military during the time he was on medical hold" he was permitted to sue for injuries sustained due to military medical malpractice); Cortez v. United States, 854 F.2d 723, 727 (5th Cir.1988) (At the time of serviceman's death due to military medical malpractice, he was on the Temporary Disability Retirement List, "was receiving disability pay and was eligible for treatment in government hospital. Neither benefit [was] sufficient to warrant a Feres bar."); Johnson v. United States, 704 F.2d 1431, 1441 n. 6 (9th Cir.1983) ($1.5 million in Veterans Act benefits deducted from serviceman's $3.5 million FTCA award); Rinelli v. United States, 706 F. Supp. 190, 194 (E.D.N.Y.1988) (serviceman on Temporary Disability Retirement List and receiving military benefits permitted to sue under FTCA for military medical malpractice); Cooper v. Perkiomen Airways, Ltd., 609 F. Supp. 969, 972 (E.D.Pa.1985) (serviceman's widow permitted to sue government for husband's death that resulted from negligence of federal air traffic controllers even though she received military compensation benefits).

Had the decision in Feres offered any other rationale that provided guidance in determining which cases should follow its own holding as opposed to Brooks, this split in authority would not have been problematic. Unfortunately, it did not.

Feres did mention two other reasons for its holding, besides the statutory benefits rationale. But neither of them was particularly helpful. First, the Court concluded that government liability in a case like Feres was not analogous to any liability of a "private individual" and, hence, was not permitted by the FTCA. See 340 U.S. at 141-42, 71 S.Ct. at 157 (FTCA did not authorize "the creation of new causes of action but [government] acceptance of liability under circumstances that would bring private liability into existence"). This explanation was highly questionable, however, and in any event, did not provide a sensible way to distinguish Brooks. Moreover, the Court abandoned it almost immediately after Feres, and has never resurrected it. See Rayonier, Inc. v. United States, 352 U.S. 315, 319, 77 S. Ct. 374, 376-77, 1 L. Ed. 2d 354 (1957) (federal government not immune under the FTCA for negligence of United States Forest Service employees); Indian Towing Co. v. United States, 350 U.S. 61, 64-65, 76 S. Ct. 122, 124-25, 100 L. Ed. 48 (1955) (rejecting claim that United States is not liable under FTCA for the "negligent performance of 'uniquely governmental functions' ").

The Court's second rationale in Feres was that all military personnel should be subject to a uniform rule governing compensation for injuries sustained while in the service. See Feres, 340 U.S. at 142-43, 71 S.Ct. at 157-58. In offering this explanation, the Court noted that "[t]he relationship between the government and members of its armed forces is 'distinctively federal in character.' " Id. at 143, 71 S.Ct. at 158. The Court further mentioned that the significant variation in tort recoveries that would inevitably result from the FTCA's lex loci provision would be inequitable in the military context. Id. at 142-43, 71 S. Ct. at 157-58.

Although this uniformity rationale was intelligible, it neither explained Feres's bar on tort suits nor accounted for the holding in Brooks. After all, if the impetus for Feres was the idea that all FTCA claims by military personnel should be controlled by a uniform federal law, then one would not have expected Feres to bar all such claims without discussion. It would have been just as plausible for the Court to have begun developing a uniform federal common law of torts--analogous, say, to admiralty--that would be applied to military claims and that would subsequently be articulated on a case-by-case basis by the lower courts.

Instead, Feres did preclude FTCA suits by military personnel, and further suggested that the government compensation system was the applicable uniform federal remedy. See 340 U.S. at 144, 71 S. Ct. at 158. This step created a logical inconsistency because this same uniform federal remedy applied to the plaintiffs in Brooks as well, and yet they were not barred by it from bringing a FTCA tort suit. Had the court chosen to create a federal common law, later courts might have distinguished Brooks from Feres on the grounds that the plaintiffs in Brooks--unlike those in Feres--were sufficiently removed from military duty so that uniformity of federal law was neither required nor, perhaps, even appropriate. Military plaintiffs would, in both instances, have tort remedies. Those whose accidents were closely linked to the government, would be covered by a uniform federal tort law. Those further removed would be bound by variable state tort rules.

Given the absence of any seemingly consistent rationale in Feres itself, it is not surprising that the Court soon developed a new, after-the-fact explanation for its holding. In United States v. Brown, the Court recharacterized Feres and said:

The peculiar and special relationship of the soldier to his superiors, the effects of the maintenance of such suits on discipline, and the extreme results that obtain if suits under the Tort Claims Act were allowed for the negligent acts committed in the course of military duty led [us] to read the Act as excluding claims of this character.

Brown, 348 U.S. at 112, 75 S.Ct. at 143.

With this new spin, the Court interpreted Feres to preclude some, but not all, actions in order to keep courts from second guessing military policies and to prevent their likely interference with military discipline. This explanation had two advantages.

First, it seemed to justify a distinction between Brooks and Feres. As the Court had already noted in Feres, "Brooks' father, riding in the same car recovered for his injuries...." 340 U.S. at 146, 71 S. Ct. at 159. This fact implied that the Brooks brothers' own FTCA recovery--given the time, place, and manner of their injuries--involved no more military second guessing or interference with discipline than did their civilian father's award.

Second, by linking Feres to discipline the Court tied into a line of influential scholarship, emerging at the time, that offered both an explanation and justification (albeit retrospective) for its otherwise extraordinarily willful refusal to follow the language and probable intent of the FTCA. Professors Alexander Bickel and Harry Wellington, for example, in a celebrated article, argued that federal courts are justified in refusing to comply with federal laws that apparently require them to behave in ways that are so inappropriate for judicial bodies as to raise structural constitutional questions. See, Alexander M. Bickel & Harry H. Wellington, Legislative Purpose and the Judicial Process: The Lincoln Mills Case, 71 Harv.L.Rev. 1 (1957). They contended that in such situations courts should find ways to avoid acting, even if such avoidance involves tortured statutory construction. Only if Congress expressly requires them to do so should courts act in such cases. Even then, it is preferable that Congress make its will known after the courts have given warning of the constitutional dangers at stake.6

In later writings, Professor Bickel, in particular, developed this theme further. See, e.g., Alexander M. Bickel, The Least Dangerous Branch (2d ed. 1986); Alexander M. Bickel, The Passive Virtues, 75 Harv.L.Rev. 40 (1961). The importance of his approach, both in this country and abroad, is obvious. See, e.g., Greene v. McElroy, 360 U.S. 474, 506-07, 79 S. Ct. 1400, 1418-19, 3 L. Ed. 2d 1377 (1959) (finding that a loyalty security program established administratively, which did not permit confrontation of witnesses, was unlawful without explicit executive or congressional authority); Kent v. Dulles, 357 U.S. 116, 128-29, 78 S. Ct. 1113, 1119-20, 2 L. Ed. 2d 1204 (1958) (Congressional delegation of power to issue passports to Secretary of State did not give the Secretary unbridled discretion to deny communists passports without explicit Congressional authorization); cf. Barenblatt v. United States, 360 U.S. 109, 137-40, 79 S. Ct. 1081, 1100, 3 L. Ed. 2d 1115 (1959) (Black, J., dissenting) ("If Congress wants ideas investigated, if it even wants them investigated in the field of education, it must be prepared to say so expressly and unequivocally.); see also Can. Const. (Constitution Act, 1982) pt. I (Canadian Charter of Rights and Freedoms), Sec. 33 (Canadian non-obstante clause providing that legislature may abrogate enumerated rights that are provisionally enforced by the Supreme Court provided that the legislature explicitly decides to do so).

Interfering with military discipline and second-guessing military policy are two activities that fit comfortably in a list of things that courts should try to avoid doing if at all possible. Perhaps for this reason, or perhaps for the want of any other seemingly coherent explanation, the "disciplinary" reading of Feres took hold and in time became dominant. See United States v. Shearer, 473 U.S. 52, 57, 105 S. Ct. 3039, 3042, 87 L. Ed. 2d 38 (1985) (stating that "Feres seems best explained" by the discipline rationale) (quoting United States v. Muniz, 374 U.S. 150, 162, 83 S. Ct. 1850, 1858, 10 L. Ed. 2d 805 (1963)). Indeed, at one point in the doctrine's recent development, the Supreme Court suggested that the other rationales--the existence of the federal compensation scheme and the desirability of a uniform federal rule--were "no longer controlling." Shearer, 473 U.S. at 58 n. 4, 105 S. Ct. at 3043 n. 4.

Unfortunately, the emphasis on discipline led the circuit courts--and perhaps the Supreme Court itself--onto a path that they were ultimately unwilling to follow. Because key questions of military discipline often seem to involve the government's control of the tortfeasor rather than the victim, the Feres analysis over time, came to focus as much on the relationship between the military and the injurer as it did on the relationship between the military and the injured plaintiff. This view of discipline is evident, if not necessarily determinative, in case after case in which courts found a Feres bar.7

But any discipline rationale that focuses on the relationship of the tortfeasor to the government has a fundamental flaw--namely, that the same acts, by the same injurer, in the same disciplinary relationship to the government, lead without question, to government FTCA liability when the victim is a civilian. See Johnson, 481 U.S. at 700, 107 S.Ct. at 2074 (Scalia, J., dissenting); Stencel Aero Engineering Corp. v. United States, 431 U.S. 666, 676-77, 97 S. Ct. 2054, 2060, 52 L. Ed. 2d 665 (1977) (Marshall, J., dissenting). Feres bars the suit only if the injured party is a member of the military. This inherent conflict was bound to surface sooner or later. Ironically, it did not emerge in a case where the claimant was a civilian and severe military discipline issues between the government and the military injurer were, nonetheless, at stake. Rather, it appeared in a case involving a member of the armed services whose estate sued the government for FTCA respondeat superior liability on account of the alleged negligence of civilians. See United States v. Johnson, 481 U.S. 681, 107 S. Ct. 2063, 95 L. Ed. 2d 648 (1987).

In Johnson, the victim was a Coast Guard helicopter pilot on a rescue mission and was killed due to a radar mishap for which the Federal Aviation Administration was responsible. Id. at 683, 107 S.Ct. at 2064-65. The decedent was on active duty, and so the government argued Feres. The Eleventh Circuit concluded, however, that since the injurer was not in the military, no issues of discipline (by now seemingly the dominant rationale for the Feres ) were at play. Id. at 684-85, 107 S.Ct. at 2065-66.

On appeal, the Supreme Court could have found that the facts raised significant discipline questions involving the decedent-servicemember and the government. It could then have held that these by then perfectly acceptable Feres grounds barred the suit. Or it could have followed Justice Scalia, who in an extraordinarily powerful dissent, argued against extending Feres to cases in which the injurer was not in the military. Id. at 692-703, 107 S.Ct. at 2069-75 (Scalia, J. dissenting).8 Either way, discipline would have remained the dominant Feres rationale. Instead, the Court did neither.

The Feres doctrine, the court said in Johnson, did not depend on discipline alone. It rested, rather, on three grounds--the desirability of a uniform federal rule, the existence of the federal compensation scheme, and concerns for military discipline. Id. at 689-91, 107 S.Ct. at 2068-69. Cf. Stencel Aero Engineering Corp. v. United States, 431 U.S. at 672-74, 97 S.Ct. at 2058-59 (considering these three factors in holding that Feres barred third party indemnity claim against government by manufacturer of a defective product that injured a serviceman while on duty). Thus, the Court resurrected its original justifications (except, of course, the totally abandoned "individual liability" analogue) and made clear that those rationales were no longer, "no longer controlling." Shearer, 473 U.S. at 58, n. 4, 105 S. Ct. at 3043, n. 4. These two original grounds, together with disciplinary considerations, both explained Feres, see Johnson 481 U.S. at 688-91, 107 S.Ct. at 2067-69, and justified barring the plaintiff's FTCA suit in Johnson itself. Id. at 691-92, 107 S.Ct. at 2069-70. No one factor governed, and all were to be evaluated.

Because the lower courts have found the rationales other than discipline extremely difficult to apply in a coherent manner, see e.g. Elliott v. United States, 13 F.3d 1555, 1559 (11th Cir.) (first two rationales "provide no help in determining when an injury occurs 'incident to service' "), vacated for reh'g in banc, 28 F.3d 1076 (11th Cir.) (in banc), affirming district court judgment by equally divided court, 37 F.3d 617 (11th Cir.1994), it is not surprising that Johnson--a decision that we are bound to follow--left both the doctrine and the lower courts more at loose ends than ever. The Fifth and Eleventh Circuits, for example, have attempted to circumvent this problem by conducting a nominal three factor Feres analysis only after resolving the threshold--and in practice determinative--issue of whether the plaintiff's injury was sustained "incident to service." See e.g. Parker v. United States, 611 F.2d 1007, 1013-15 (5th Cir.1980) (adopting a three-part test for deciding whether servicemember's activity is "incident to service"); Pierce v. United States, 813 F.2d 349, 352-54 (11th Cir.1987) (applying Parker test). This Circuit, on the other hand, has instructed the district courts to analyze servicemember claims "in light of the 'three broad rationales' underlying Feres, and to determine whether [the plaintiff's] injuries 'arise out of or in the course of activity incident to service.' " Sanchez v. United States, 839 F.2d 40, 42 (2d Cir.1988) (per curiam), but has given no guidance as to how this is to be done in a coherent fashion. Neither approach is fully satisfactory.

It seems to us that there are two ways in which the Feres doctrine can be dealt with today. They would both lead to the same result in this case. The first approach would be to agree with Justice Scalia's dissent in Johnson and to admit that Feres is a mistake--was perhaps always a mistake--and should be scrapped. That option is, of course, not open to us.9 After all, this case does not present one of those exceedingly rare situations in which a lower court can so clearly foresee that the Supreme Court will reverse itself that it can ignore a Supreme Court precedent. See Barnette v. W.Va. State Bd. of Educ., 47 F. Supp. 251, 252-53 (S.D.W.Va.1942) (Parker, J.) (disregarding prior Supreme Court flag salute precedent where a majority of the Court had expressed its desire to overrule its prior decision and the threatened violation of religious liberty was clear), aff'd, 319 U.S. 624, 63 S. Ct. 1178, 87 L. Ed. 1628 (1943). Feres has been regularly--albeit inconsistently--applied by the Supreme Court, and the mere existence of a convincing dissent in a closely divided Court cannot alter its binding force.

The second possibility is to go back to the Feres and Brooks cases and see whether, taken together, these decisions provide a consistent thread of meaning that we can use to sew up the holes that exist in the doctrine today. If such a meaning exists, and if that meaning is consistent with the three "factors" enumerated in Johnson, then these factors can function as guideposts for the lower courts--rather than serve merely as ritual words to be announced, formulaically and ex cathedra, after a court has decided to apply or to eschew Feres on some other ground. We think such a reading is possible and that it results in a workable doctrine.10

First came Brooks, which, despite the existence of a federal statutory scheme of military death and disability benefits, permitted members of the armed services to recover under the FTCA so long as their "injuries [were] not caused by their service except in the sense that all human events depend upon what has already transpired." Brooks, 337 U.S. at 52, 69 S.Ct. at 920. Then came Feres, which strongly intimated that the existence of a federal compensation scheme should bar all suits against the United States government by military personnel, just as workers' compensation laws regularly barred injured employees from suing their employers. See Feres, 340 U.S. at 144-45, 71 S.Ct. at 158-59. As we have already noted, the Court's earlier result in Brooks--reaffirmed in later cases--has always seemed out of line with Feres. In fact, however, this seeming inconsistency is more apparent than real.

There is a fundamental difference between the government's compensation scheme for its military "employees" and workers' compensation laws. The government's compensation scheme is not limited to compensating military "employees" who are injured in ways that "arise out of or in the scope of employment." It covers servicemembers' injuries, as well as diseases, that are completely unrelated to the military enterprise. See, e.g., 38 C.F.R. Sec. 3.1(m) (regulation defining "in line of duty" coverage as coverage for "injury or disease incurred or aggravated during a period of active military, naval, or air service unless such injury or disease was a result of the veteran's own willful misconduct."). Workers' compensation laws, instead, only compensate employees for injuries that arise out of or in the course of their employment. And under these laws, it is only this limited class of injuries that give rise to a Feres-like prohibition of suits by employees against their employers.

A helpful way of understanding the significance of this difference is to consider the military's death and disability benefits as essentially two different plans combined into one statutory scheme. The first component is analogous to a workers' compensation system, and covers military employees who are injured in ways that arise out of or in the scope of their employment. The second component is like a supplemental health and disability plan that an employer voluntarily provides to its employees. This latter "plan" covers "employees" regardless of how or when they are injured, and, specifically, whether or not the injury arises out of the employer's enterprise.11

No one has ever imagined that the existence of a voluntary benefits plan for private employees means that such employees, if they should be injured outside the scope of their employment (but in ways covered by the voluntary plan), may not sue their employers in tort.12 Since the compensation provided under such voluntary plans is not a function of workers' compensation laws (because the injuries are not work related ), tort actions are not barred.13

Brooks and Feres make the same distinction. Feres bars suits where compensation is given under a military analogue to workers' compensation. Brooks allows suits when compensation occurs, for non-work related injuries, i.e., those that would not be covered by workers' compensation. Read this way, Brooks, Feres and their respective progeny are not only substantially consistent with each other, but they also achieve a rough parity with workers' compensation laws. That is, thus understood, they accomplish the very result that the Court sought in Feres.

Whether this desire to make the FTCA conform with workers' compensation laws was wise, and whether, in any event, it was too willful is, of course, not for us to say. Similarly, it is not for us to say whether Feres, even if wise and appropriate when decided, yields appropriate or sensible results today. We can note, however, that both the military and the workers' compensation schemes have fallen significantly out of line with ordinary tort recoveries14 and, thus treat injured employees unlike other injured parties. We can speculate on whether this fact undermines the original intent of such laws to treat injured employees at least as well as other injured parties. See, e.g., Andrea Giampetro-Meyer & Ann M. Balcerzak, Renegotiating the Bargain: An Analysis and Evaluation of Alternatives for Revising the Exclusive Remedy Provision in Maryland's Workers' Compensation Act, 21 U.Balt.L.Rev. 51, 54-55 (1991) (noting that workers' compensation laws were enacted in response to employee-obstacles to recovery for job-related injuries) (hereinafter, "Renegotiating the Bargain "); Prosser & Keeton, Sec. 80 at 572-74 (brief history of workers' compensation legislation). Finally, we may notice that the increasing difference in tort and workers' compensation recoveries has led many state courts to find creative (and perhaps wilful) ways around the traditional statutory bar on employee-employer lawsuits. Notably, if this judicial trend toward permitting employee "end-runs" around the workers' compensation laws continues, see generally, Renegotiating the Bargain, 21 U.Balt.L.Rev. at 57-58; Brad A. Elward, Comment, The Interplay Between Contribution and Workers' Compensation in Illinois: Putting An End to Backdoor Recoveries, 13 S. Ill. U. L.J. 103 (1988), civilian employees will receive greater benefits under porous state statutory schemes than will military employees under the Feres doctrine that ironically, itself, had stretched the FTCA in order to achieve rough parity between military and civilian employees.15

But such considerations, if they are relevant at all, are for the Supreme Court to ponder. For us it is enough to say that the Feres doctrine remains the law, and that the distinction between Feres, Brooks, and the decisions that followed each, can be understood in the light of Feres's original objectives. Viewed in this way, the distinction becomes an effective guide for district courts adjudicating these types of cases. This distinction, moreover, withstands the after-added factor of military discipline. When used with appropriate moderation, see Johnson, 481 U.S. at 688-91, 107 S. Ct. at 2067-69, discipline can add its own unique flavor to the Feres sauce without destroying the original taste.

Even if Feres were not justified by a desire to emulate workers' compensation laws in the military context, we believe that the doctrine could still find a basis in the prudential grounds raised by Professors Bickel and Wellington. In other words, some FTCA suits by military personnel would still be barred because their prosecution would lead to significant judicial interference in military decisions. But what is "significant interference?" In answering this question we must revisit the distinction between discipline issues affecting the injurer and discipline issues affecting the injured military-plaintiff.

Considerations of military discipline that stem from the relationship between the injured servicemember and the military would, at first glance, appear to be quite significant. These types of situations frequently raise issues that many feel the courts should best avoid, such as

[t]he peculiar and special relationship of the soldier to his superiors, the effects of the maintenance of suits on discipline, and the extreme results that might obtain if suits under the Tort Claims Act were allowed for negligent orders given or negligent acts committed in the course of military duty.

Brown, 348 U.S. at 112, 75 S.Ct. at 143.

But in fact, examining a claim's "disciplinary" ramifications as a separate Feres factor usually does not add very much. It is like adding a handkerchief to a blanket. As a practical matter, in most such cases the military plaintiff would be barred from suing anyway because any claims that raise disciplinary issues relating to the plaintiff almost always fit within the classic workers' compensation paradigm--that is, they arise out of or in the course of the claimant's military employment. In other words, Feres's bar on claims "aris[ing] out of or in the course of activity incident to service," 340 U.S. at 146, 71 S. Ct. at 159, understood in the light of its workers' compensation rationale, will almost inevitably keep the judiciary out of matters of military discipline. This is not to say that there aren't cases that raise many questions of military policy. But most, if not all claims that could potentially lead to inappropriate court meddling, would be barred in any event because they are also employment-related.

The disciplinary relationship between the government and the injurer is very different. Most of the time it is immaterial for three reasons: the holding in Johnson; the fact that Feres does not bar civilian plaintiffs; and the nature of tort damage awards.

The first reason is clearly not dispositive. It neither supports nor undermines a Feres bar, but rather simply points out that the relationship between the injurer and the defendant is not an essential part of the Feres analysis. Johnson tells us that the Feres doctrine can bar a servicemember's claim even if the injurer is not a military employee and, thus, no issues of discipline exist as to the injurer. See Johnson, 481 U.S. at 686, 107 S.Ct. at 2066 ("this Court has never suggested that the military status of the tortfeasor is crucial to the application of the doctrine"). This may be because the discipline relationship between the military-plaintiff and the government is already so great that a Feres bar is warranted in any event. Or it may rest on one of the original Feres grounds, such as the workers' compensation analogy.16

One might conclude from this that the military-injurer discipline relationship is irrelevant to Feres. But since all Johnson actually says is that a Feres bar may exist absent a military-injurer, the most that it can mean is that a discipline relationship between the military and the injurer is not a necessary condition for Feres to apply. Johnson obviously does not mean that the existence of such a strong discipline relationship could never justify a Feres result.

The second reason is more compelling. Feres does not bar suits against the government when the injured plaintiff is a civilian. This remains the case even though the injurer is in the military and military discipline is directly involved. See e.g. Sheridan v. United States, 487 U.S. 392, 401, 108 S. Ct. 2449, 2455, 101 L. Ed. 2d 352 (1988) (injured civilian shot by drunken serviceman could sue government under FTCA for navy personnel's negligence in knowingly allowing drunken serviceman to leave navy base with rifle); Doggett, 875 F.2d at 689-95 (government may be held liable under the FTCA for injuries sustained by a civilian as a result of Navy's negligent supervision of drunken personnel); Kohn v. United States, 680 F.2d 922, 926 (2d Cir.1982) ( Feres--though it precluded recovery by the estate of a murdered serviceman--did not bar claims by his bereaved parents for the Army's alleged infliction of emotional distress on them subsequent to their son's death); but cf. Daberkow v. United States, 581 F.2d 785, 788 (9th Cir.1978) (West German air force pilot killed while conducting joint training mission with U.S. Air Force barred by Feres from suing United States Government for alleged Air Force negligence).

Accordingly, in any number of civilian cases, the alleged judicial inquiry into (and interference with) military affairs, occurs anyway. And if this interference occurs regularly in any event, it cannot possibly raise the constitutional concerns that warrant tortured statutory construction and judicial abstention. Since this type of alleged interference is part and parcel of ordinary court behavior when the plaintiff is not in the military, it is hard to argue that such routine judicial inquiry must be avoided at all costs simply because plaintiff is in the military.

The third and final reason is equally strong. It is difficult to see how FTCA damage awards can, except in the rarest of cases, interfere with a disciplinary relationship between the government and the military tortfeasor. The issue in these cases is typically not whether the military is permitted to do certain things. It is, instead, whether a member of the armed services has behaved negligently or otherwise wrongfully. If that employee committed a tort--did something that if done by a civilian would give rise to liability--and the act was in the scope of government employment, the government is prima facie liable unless something about the injured plaintiff--like military ties--bars liability. If the government is liable the government is held to pay damages. But paying damages does not mean that the military is told by a court that it must do things differently, or even that it must take steps to control its employees.

Injunctions and regulations tell people what they must do and what they must not do, and it is these types of intrusions that would entangle courts in military affairs. Tort judgments do neither of these things. Of course, the military or other agencies of government may themselves decide to alter some forms of military behavior as a result of a damage award. But that decision, that interference, is not the court's. Pursuant to the FTCA, courts merely determine whether analogous behavior by a private-sector employee would give rise to some form of fault-based vicarious liability on the part of a private-sector employer. And under the FTCA, courts simply hold that similar harms done by military employees of the government are compensable costs of the military enterprise. Whether or not a particular cost is worth incurring is a decision that the military must make. The dynamics of that subsequent decision, moreover, hardly give rise to the prudential concerns regarding the constitutional separation of powers and the integrity of Article III courts that properly worried Bickel and Wellington. See Lincoln Mills, supra.

From all of this, one might expect that disciplinary considerations regarding the injurer's military relationship to the government are usually ignored in Feres cases. Instead, many courts--perhaps misled by the Supreme Court's temporary preoccupation with discipline--have examined the military relationship ostensibly to avoid judicial interference in military matters. Indeed, our own prior decisions have, at times, engaged in just such an analysis. See e.g. Sanchez v. United States, 878 F.2d 633, 638 (2d Cir.1989) (considering the need for Marine Corps officers to testify regarding their policies in hiring allegedly negligent auto repair mechanics); Bozeman v. United States, 780 F.2d 198, 202 (2nd Cir.1985) (considering the need for Army officers to testify regarding their policies in staffing and operating a non-commissioned officers' club).

In all these cases, however, there were other grounds that justified a Feres bar--apart from any discipline relationship between the government and injurer. Either the first two Feres factors or, at times, a crucial discipline relationship between the injured plaintiff and the military, were present and sufficiently explained why Feres applied. See Sanchez, 878 F.2d at 637 ("facility at which the alleged negligence occurred was open only to a limited class of military or military related patrons" and thus was an incident of plaintiff's military employment); Bozeman, 780 F.2d at 201 (plaintiff's decedent "was only entitled to be in the NCO club because he had the appropriate rank, was a member of the Army and was on active duty status" and thus his drinking fell squarely within the scope of his employment). And for the doctrine to survive in a coherent way, we conclude that the district courts would do best to look primarily to these other grounds--just as the Supreme Court did in both Brooks and Feres.

Still, there are some very rare cases in which injurer discipline is crucial, and Feres, as it has developed, has left room for them. The doctrine provides a safeguard against the odd case in which truly significant issues of discipline are at stake, and in which, for unusual reasons, the other factors do not result in a Feres bar. In such rare and remarkable cases a discipline relationship, whether between the government and the plaintiff or even between the government and the injurer, is by itself enough to warrant a Feres bar.

In Shearer the Supreme Court applied Feres to bar a mother's FTCA suit for the kidnapping and murder of her son, an off-duty soldier who was off base at the time of his abduction. See 473 U.S. at 53-54, 105 S.Ct. at 3040-41. The mother complained that the Army "failed to exert a reasonably sufficient control over" her son's assailant, a soldier who had a prior record of violent crime. Id. Concluding that the mother's claim struck "at the core" of military discipline, id. at 58, 105 S.Ct. at 3043 ("This allegation goes directly to the 'management' of the military; it calls into question basic choices about discipline, supervision, and control of a serviceman"), the Court invoked the Feres doctrine. The decision in Shearer emphasized

[t]hat to permit this type of suit would mean that commanding officers would have to stand prepared to convince a civilian court of the wisdom of a wide range of military and disciplinary decisions; for example, whether to overlook a particular incident or episode, whether to discharge a serviceman, and whether and how to place restraints on a soldier's off-base conduct.

Id.

Significantly, the Supreme Court specifically distinguished "the negligence alleged in the operation of a vehicle" from the type of claim that gave rise to the serious disciplinary issues that concerned it in Shearer. Id. Thus Shearer seems properly to represent the rare case that warrants a free standing "disciplinary" basis for a Feres bar. As such, it correctly defined itself as the exception rather than the rule.17

We do not doubt that "safeguarding the integrity of military discipline" can be an appropriate basis for justifying, or at least rationalizing, the Court's construction of the FTCA in Feres. And if the Supreme Court had stuck to its guns after Shearer, and continued to exclude all other factors, we would have been quite "disciplined" in our subsequent approach. But after Johnson, which reaffirmed the desirability of a uniform federal rule and the existence of a statutory compensation system as proper Feres considerations, we feel more at liberty to examine the relative importance of each factor and the necessary interplay between them. That examination reveals that discipline considerations, crucial as they may be, need not play an independent role in the majority of cases. Where discipline issues involve the plaintiff and the military, their demands are usually already met by complying with the other factors. Where the discipline issues involve the injurer and the military, they are only relevant in the extreme case, like Shearer, as Shearer itself suggests.

Under Brooks, Feres, Stencel, Shearer, and Johnson, an appropriate test for applying the Feres doctrine must respect: (1) the Supreme Court's stated concern for keeping courts away from delicate questions involving military discipline; (2) Feres's clear intention to replace the contingencies of local tort law with a uniform federal scheme; and (3) Feres's original desire that this uniformity is to be achieved through exclusive recourse to the federal system of military death and disability benefits. We believe that these concerns are, in fact, fully captured by the original language in Feres that barred suits by military claimants for all injuries "aris[ing] out of or in the course of activity incident to service." 340 U.S. at 146, 71 S. Ct. at 159.

As we have noted, this language derives from a familiar phrase ("arising out of or in the course of employment"), which has a well defined meaning in the context of workers' compensation. This definition, moreover, is closely related to the scope of the government's vicarious liability--since workers' compensation and respondeat superior are both concerned with charging costs to the business enterprise that can be fairly said to engender them. See Childers, 190 Cal.App.3d at 801, 235 Cal. Rptr. at 644. Thus, the two issues in this case coalesce. Because respondeat superior requires the government to pay for third-party injuries that are foreseeable costs of the general military enterprise, and because the federal statutory compensation scheme, buttressed by the Feres doctrine, requires the government to pay for all employment-related injuries that are sustained by servicemembers only through a system of workers' compensation payments,18 it is appropriate that courts interpret the test for both in similar ways.19

We conclude that the same parameters of employment-related conduct that determine the existence of governmental vicarious liability should also generally define the limits of the Feres doctrine. In deciding whether Feres applies to bar a particular claim, a court should consider whether, at the time of the accident, the military-plaintiff's activity would have imposed respondeat superior liability on the government had that activity injured a civilian. If the answer is "yes," then the Feres doctrine applies because the servicemember sustained his or her injuries "in the course of activity incident to service." Conversely, if the answer is "no," there should be no Feres bar (absent a truly unusual discipline problem)--since the accident occurred outside the scope of military employment and, thus, would not be covered by a normal workers' compensation scheme.20

In the case before us, we hold that the link between Taber's activity when he was injured and his military status is too frail to support a Feres bar. Put another way, given the circumstances surrounding Taber's collision with Maine, it is hard to imagine that the government would have been vicariously liable if Taber had driven negligently and injured a third party. If Maine's actions fit the traditional pattern of a "detour," which gives rise to employer liability even where the employee's activity does not directly benefit the employer, so Taber's actions represent the classic "frolic," which describes behavior that is not properly part of the employer's enterprise. See Prosser & Keeton, Sec. 70 at 503-05; Young B. Smith, Frolic and Detour, 23 Col.L.Rev. 444, 716 (1923).

There is nothing characteristically military about an employee who, after working-hours are done, goes off to spend a romantic weekend with a companion. Nor is there anything particularly military about having dinner with that companion's family at their home, and helping to fix their car. Finally, there is nothing especially military about returning to the companion's house intending to spend the rest of the weekend engaged in more intimate rest and recreation.

The accident that followed, on the open road and on the way to Stills's house had "nothing to do with" Taber's military career and was "not caused by service except in the sense that all human events depend upon what has already transpired." See Brooks, 337 U.S. at 52, 69 S.Ct. at 920.21 The only difference between Brooks and this case is that the Brooks brothers were on furlough, while Taber was on liberty, and that in Brooks the civilian riding in the car was the Brooks's father, while here it was Taber's girlfriend. There are no Supreme Court or Second Circuit decisions that require us to give significant weight to this difference, or to apply the Feres bar on any other grounds.

In Kohn, 680 F.2d at 925 and in Camassar v. United States, 531 F.2d 1149, 1151 (2d Cir.1976) (per curiam), we held that Feres barred plaintiffs' suits. In both of these cases the injuries occurred on military bases, and the injured servicemembers were engaged in activities that would ordinarily be covered by workers' compensation statutes. See Kohn, 680 F.2d at 924 (plaintiff's decedent was shot to death by military colleague while both were assigned to a military police narcotics unit); Camassar, 531 F.2d at 1150 (decedent navy petty officer killed while he was leaving ship for authorized liberty).

In Bozeman, the claimant asked us to rule that because the deceased servicemember was off duty, on liberty and off base when he was killed in an auto accident, Feres was inapplicable. See 780 F.2d at 200. In response, we noted that although these facts were considerations in our Feres analysis, none was determinative. In applying Feres, we made much of the fact that the servicemember and the tortfeasor (his cohort) got drunk at the NCO club and that the servicemember was "only entitled to be in the NCO club because he had an appropriate rank, was a member of the Army, and was on active duty status." 780 F.2d at 201. Had Bozeman injured a third party, it would follow from holdings like Bushey, and our own today, that the government would be vicariously liable. Small wonder then that we found Bozeman's link to the military enterprise sufficient to support a Feres bar. But we did this despite the fact that Bozeman was on liberty and the accident occurred "off base."22

In Sanchez v. United States, 878 F.2d 633 (2d Cir.1989), we applied Feres in a case in which the plaintiff had been injured while riding in a car that had been negligently repaired at a military mechanic's shop. Id. at 634. In coming to our decision, we placed particular importance upon the fact that the Marine Corps garage, whose negligence was allegedly responsible for the accident,

was not open to the public but only to military personnel and certain civilians connected with the base, in recognition of the fact that "[t]he automobile plays a highly vital role in the life of the military patron and his of her family."

Id. at 637 (citation omitted).

Although we think the facts in Sanchez are at the very fringe of the doctrine's applicability--beyond which we are unwilling to go--we note that the plaintiff's unique, and indeed necessary connection to the mechanic's shop was sufficiently incidental to his military service to invoke the Feres bar.

Other than the naked fact that Taber was in the Navy at the time of his injury, there is no government/plaintiff relationship of any significance in this case. Arguably, there is some government/tortfeasor relationship that might entail limited disciplinary concerns, but these are neither of the magnitude nor of the type that concerned the Court in Shearer. And we conclude that the absence of any government/plaintiff disciplinary issues, together with the fact that Taber's injury occurred in a private car, on an open road and while Taber was on liberty, outweighs whatever significance the negligible presence of a government/tortfeasor relationship might still have after Johnson.

It is true that Taber is covered by the government compensation system as were the plaintiffs in Brooks and in Brown. But as in those cases, Taber's coverage is akin to voluntary employee health and injury benefits that accrue independently of the nature of an employee's injury. As such, they are very different from the type of coverage that is required under workers' compensation schemes. Therefore, as in Brooks and Brown, this type of voluntary coverage will merely serve as a set-off against any FTCA award; it does not bar an FTCA action. See Brown, 348 U.S. at 111, 75 S.Ct. at 142-43; Brooks, 337 U.S. at 53-54, 69 S.Ct. at 920-21.

For all these reasons, we hold that Feres does not apply here.III. Liability and Damages

The district court found Maine liable to Taber and assessed damages in the amount of $300,000.00. The government, having been exempted from liability under the district court's respondeat superior decision, did not participate at trial and had no opportunity to contest these findings. Maine represented himself in these proceedings. At oral argument, the government's very able attorney requested that, should we determine that the government was vicariously responsible for Maine's actions, and that Taber's suit was not barred by Feres, we vacate that judgment and permit the government to contest both Maine's culpability and Taber's damages.

The government's request apparently stems from its concern that, upon remand, it will be collaterally estopped from disputing either Maine's negligence (on the basis of which the government's liability depends) or the district court's previous calculation of damages. The government undoubtedly raises important questions. None of them, however, has been fully briefed or argued by the parties. We therefore leave entirely to the district court what, if any, preclusive effect the judgment against Maine will have on the government's defense against Taber's assertion of liability under the FTCA.

We further note that Taber currently holds a valid judgment against Maine, from which Maine has not appealed. Had Maine sought review of that judgment, we might have considered whether the Federal Employees Liability Reform and Tort Compensation Act, 28 U.S.C. Sec. 2679 (providing a statutory immunity for federal employees whose employment-related conduct injures others), shielded him from personal liability. Although Maine was not certified by the Attorney General to have been acting within the scope of his employment in accordance with Sec. 2679(d)(1), subparagraph (d)(3) provides that where the Attorney general has failed to certify,

the employee may at any time before trial petition the court to find and certify that the employee was acting within the scope of his office or employment ... [and that upon] such certification by the court, such action ... shall be deemed to be an action ... against the United States ..., and the United States shall be substituted as the party defendant.

28 U.S.C. Sec. 2679(d)(3). In opposition to the government's motion for summary judgment, Maine did submit an affidavit which set forth the undisputed facts upon which we base our present ruling that he acted within the scope of his employment when he injured Taber.

We have previously suggested that substantial compliance with section 2679's procedural conditions may, under certain circumstances, invoke the statutory immunity. See B & A Marine Co. v. American Foreign Shipping Co., 23 F.3d 709, 715-16 & n. 4 (2d Cir.) (defendants' brief, which requested the district court to find that they acted within the scope of their employment, and district court's finding to that effect, "sufficiently complied with the certification requirement"), cert. denied, --- U.S. ----, 115 S. Ct. 421, 130 L. Ed. 2d 336 (1994). It is therefore possible that Maine's affidavit in opposition to summary judgment could be construed to be a "petition" for certification under the statute. See id. at 715 n. 4. And if Maine had appealed from the district court's final judgment, our respondeat superior ruling could be taken to satisfy section 2679's judicial certification requirement. Because he did not seek appellate review, however, we cannot reach the issue.

For the reasons stated above, we reverse the district court's judgment and remand the case to the district court for further proceedings consistent with this opinion.

I concur in the judgment.