Linda S. Hamilton appeals a judgment of the district court in a diversity action that she is civilly liable, as a joint venturer and coconspirator, for the killing of Michael Halberstam by Bernard C. Welch, Jr. on December 5, 1980, in the District of Columbia. The appellee, Elliott Jones Halberstam, is the personal representative of the estate of Michael Halberstam, her late husband. She brought this wrongful death and survival action1 for damages on behalf of the estate, Michael Halberstam's two children, and herself. Halberstam alleged that Bernard Welch and Linda Hamilton engaged in a joint criminal venture and conspiracy in the course of which Welch killed Michael Halberstam while burglarizing the Halberstams' home. Welch failed to file an answer, and the district court entered a default judgment against him on May 19, 1981. Hamilton actively defended the suit, but after a nonjury trial on January 12, 1982, the court found her jointly and severally liable with Welch and entered a judgment against both in the amount of $5,715,188.05. Hamilton now appeals the issue of her liability.2 We conclude that the district court's findings of fact are not clearly erroneous, and that it applied the proper law; accordingly, we affirm the judgment against Hamilton.

This case arises out of the shocking climax to a coldly efficient criminal campaign that had confounded, frustrated, and ultimately terrorized the Washington area. We are asked to determine the civil liability of the passive but compliant partner to this rampage that left widowed the wife of one of the community's most eminent physicians. As a result of Welch's innumerable burglaries over the course of five years, he and Hamilton acquired a fortune that would have been the envy of a Barbary brigand. In the words of the district court, Hamilton:

knew full well the purpose of [Welch's] evening forays and the means by which she and Welch had risen from "rags to riches" in a relatively short period of time. She closed neither her eyes nor her pocketbook to the reality of the life she and Welch were living. She was compliant, but neither dumb nor duped, so long as her personal comfort and fortune were assured. She was a willing partner in his criminal activities.

Halberstam v. Welch, No. 81-0903, mem. op. at 5 (D.D.C. Mar. 24, 1982) [hereinafter cited as District Court Opinion]. The district court based this conclusion largely on Hamilton's own testimony.

Hamilton first met Welch in October 1975, when Welch walked up to her in an apartment parking lot and asked her for a date. Hamilton stated that this was the first and only time she saw Welch with a gun. At the time of their meeting, Hamilton, a twenty-five-year-old high school graduate, worked as a secretary-compositor at the National Academy of Sciences. Welch told Hamilton that he bought estates and invested in coins, jewelry stores, and real estate. Welch moved into Hamilton's apartment a few weeks after their first meeting; apparently, his only assets at that time were a new Monte Carlo automobile, some clothing, a watch, pocket change, and some gold coins. They continued to live together, in various residences, until Welch shot Halberstam during the course of a burglary of Halberstam's home on December 5, 1980. Trial Transcript ("Tr.") at 12-15, 29-30; Hamilton Deposition at 4-5.

In 1976, Hamilton and Welch moved to a rented house in Falls Church, Virginia. Hamilton, still employed at this time, gave Welch her salary in cash to invest for her in gold coins. Welch had no outside employment, and spent most days at home managing investments. He would leave the house four or five times each week between 5:00 p.m. and 5:30 p.m., and return between 9:00 p.m. and 9:30 p.m. This routine continued throughout the five years Hamilton lived with Welch.3 Hamilton stated that she never accompanied Welch on these evening expeditions. She did not think his absences peculiar, and said she never had a full discussion with him about where he had been. She assumed he was checking on his investments or meeting with coin and jewelry dealers. Hamilton remembered only one instance, in Minnesota, when she joined Welch on a visit to a coin dealer. Now and then Hamilton picked up coins for Welch from dealers, paying for the coins with cash Welch supplied. Tr. at 13-15, 17-18, 31; Hamilton Deposition at 6, 13, 23.

Soon Welch's "investments" bore fruit. In April 1978, after Hamilton gave birth to the first of their three children, Welch and Hamilton purchased a house in Minnesota for $102,000. Welch contributed about $55,000 in cash, and Hamilton put up about $20,000. The house was titled in Hamilton's name. They lived in Minnesota during portions of 1978, 1979, and 1980. Tr. at 22, 32-36, 41, 116; Hamilton Deposition at 23-24.

In 1979, the couple built a home in Great Falls, Virginia, valued at $1,000,000. Except for a trip to Minnesota during the summer, they lived in Great Falls from November 1979 until December 1980, when Welch was arrested for killing Halberstam. Hamilton's niece and the niece's child moved in with them. As suited their lifestyle, Welch and Hamilton bought two 1980 Mercedes-Benz cars and a station wagon, and hired a housekeeper. Tr. at 8, 18-19, 116; Hamilton Deposition at 12, 15; Plaintiff's exhibit 8.

Meanwhile, a different kind of refinement was taking place in the garage. With Hamilton's knowledge, Welch installed a smelting furnace in the garage and used it to melt gold and silver into bars. He then sold the ingots to refiners in other states. Hamilton typed transmittal letters for these sales. She also kept inventories of antiques sold, and in general did the secretarial work for Welch's "business." The buyers of Welch's goods made their checks payable to her, and she deposited them in her own bank accounts. She kept the records on these asymetrical transactions--which included payments coming in from buyers, but no money going out to the sellers from whom Welch had supposedly bought the goods. Hamilton remembered no mail from dealers in antiques or precious metals. Tr. at 24-25, 42-43, 72-83, 119-22, 126-29; Hamilton Deposition at 6, 9, 14-15, 18-21.

Not surprisingly, given the "low" cost of Welch's materials, his business was a profitable one. By 1978 Hamilton and Welch had a gross annual income in excess of $1,000,000. Hamilton's individual tax returns for 1978 and 1979 reported gross earnings of $647,569.21 and $491,762.16, respectively, from the sale of gold and silver. She took deductions, per Welch's instructions, for "cost of goods sold and/or operations" in 1978 and 1979 of $498,770.87 and $360,000, respectively--despite the absence of any evidence of payouts for such goods. Hamilton assumed that Welch filed a separate tax return. Tr. at 22-24, 38-41; Plaintiff's exhibits 6 and 7.

After the police apprehended Welch, they obtained a search warrant for the Great Falls house and discovered Welch's basement "inventory": some fifty boxes containing approximately three thousand stolen items--antiques, furs, jewelry, silverware, and various household and personal effects. While Hamilton admitted having seen the boxes, she claimed not to have seen their contents before. She said she did not go down to the basement often, although she had free access to it. Indeed, she provided policemen the key to Welch's locked basement study.4 Tr. at 8-10, 15-16, 21-22, 95-96; Hamilton Deposition at 9-10, 18-19; Plaintiff's exhibits 4 and 5.

The district court concluded that this loot "was the source of [Welch's] and Hamilton's livelihood, income and investments. Disposing of the loot was the principal business in which Welch and Hamilton engaged while at home." District Court Opinion at 5.

The primary issues raised by this appeal are what kind of activities of a secondary defendant (Hamilton) will establish vicarious liability for tortious conduct (burglaries) by the primary wrongdoer (Welch), and to what extent will the secondary defendant be liable for another tortious act (murder) committed by the primary tortfeasor while pursuing the underlying tortious activity. To illuminate these issues, we need to clarify basic elements of vicarious liability in tort and to analyze case law to see what evidence is sufficient to establish them.

Various theories of civil liability are untidily grouped under the general heading of concerted tortious action.5 To guide our deliberations, we initially outline a framework for this area of tort law, focusing on the two theories of liability the district court found applicable here--civil conspiracy, and aiding and abetting. We discuss a series of cases in the context of the framework, beginning with the sparse District of Columbia precedent. Then, after testing the district court's factual inferences against the "clearly erroneous" standard of review, we finally apply the law to these facts to determine Hamilton's liability for Michael Halberstam's death.

Prosser notes that "[t]he original meaning of 'joint tort' was that of vicarious liability for concerted action. All persons who acted in concert to commit a trespass, in pursuance of a common design, were held liable for the entire result." W. Prosser, Law of Torts Sec. 46, at 291 (4th ed. 1971). His illustration portrays a standard situation that involved this "joint tort": combined action by tortfeasors on the scene together--"one might have battered the plaintiff, while another imprisoned him, and a third stole his silver buttons." Id. (footnote omitted). Each was responsible for the others' actions.

Over time, courts applied the principle of vicarious liability for concerted action to less obvious situations, covering tortfeasors whose relationship was more subtle than Prosser's "highwaymen." The two variations significant here are (1) conspiracy, or concerted action by agreement, and (2) aiding-abetting, or concerted action by substantial assistance. These two bases of liability correspond generally to the first two subsections in the Restatement (Second) of Torts Sec. 876 (1979) [hereinafter cited as Restatement ] on "Persons Acting in Concert":

For harm resulting to a third person from the tortious conduct of another, one is subject to liability if he

(a) does a tortious act in concert with the other or pursuant to a common design with him [conspiracy],6 or

(b) knows that the other's conduct constitutes a breach of duty and gives substantial assistance or encouragement to the other so to conduct himself [aiding-abetting], or

(c) gives substantial assistance to the other in accomplishing a tortious result and his own conduct, separately considered, constitutes a breach of duty to the third person.

(Emphasis added.) See Pharo v. Smith, 621 F.2d 656, 669 (5th Cir.1980); Payton v. Abbott Labs, 512 F.Supp. 1031, 1035 (D.Mass.1981).

As pristine legal concepts, conspiracy and aiding-abetting can be distinguished clearly enough. A list of the separate elements of civil conspiracy includes: (1) an agreement between two or more persons; (2) to participate in an unlawful act, or a lawful act in an unlawful manner; (3) an injury caused by an unlawful overt act performed by one of the parties to the agreement; (4) which overt act was done pursuant to and in furtherance of the common scheme. See, e.g., Ryan v. Eli Lilly & Co., 514 F.Supp. 1004, 1012 (D.S.C.1981).

The element of agreement is a key distinguishing factor for a civil conspiracy action. Proof of a tacit, as opposed to explicit, understanding is sufficient to show agreement. See Prosser, supra, at 292; 16 Am.Jur.2d Conspiracy Sec. 68 (1979). But the agreement in a civil conspiracy does not assume the same importance it does in a criminal action. To establish liability, the plaintiff also must prove that an unlawful overt act produced an injury and damages.7 "It is only where means are employed, or purposes are accomplished, which are themselves tortious, that the conspirators who have not acted but have promoted the act will be held liable." Prosser, supra, at 293 (footnotes omitted).

Aiding-abetting includes the following elements: (1) the party whom the defendant aids must perform a wrongful act that causes an injury; (2) the defendant must be generally aware of his role as part of an overall illegal or tortious activity at the time that he provides the assistance; (3) the defendant must knowingly and substantially assist the principal violation. See, e.g., Investors Research Corp. v. SEC, 628 F.2d 168, 178 (D.C.Cir.), cert. denied, 449 U.S. 919, 101 S.Ct. 317, 66 L.Ed.2d 146 (1980); Woodward v. Metro Bank of Dallas, 522 F.2d 84, 94-95 (5th Cir.1975); Landy v. Federal Deposit Insurance Corp., 486 F.2d 139, 162-63 (3d Cir.1973), cert. denied, 416 U.S. 960, 94 S.Ct. 1979, 40 L.Ed.2d 312 (1974).8 In the "Comment on Clause (b)," the authors of the Restatement summarize these elements and explain why they create liability: "Advice or encouragement to act operates as a moral support to a tortfeasor and if the act encouraged is known to be tortious it has the same effect upon the liability of the adviser as participation or physical assistance." Id. Sec. 876, comment d.

In practice, liability for aiding-abetting often turns on how much encouragement or assistance is substantial enough. The Restatement suggests five factors in making this determination: "the nature of the act encouraged, the amount of assistance given by the defendant, his presence or absence at the time of the tort, his relation to the other [tortfeasor] and his state of mind." Id.

The prime distinction between civil conspiracies and aiding-abetting is that a conspiracy involves an agreement to participate in a wrongful activity. Aiding-abetting focuses on whether a defendant knowingly gave "substantial assistance" to someone who performed wrongful conduct, not on whether the defendant agreed to join the wrongful conduct.

Courts and commentators have frequently blurred the distinction between the two theories of concerted liability.9 Most commonly, courts have relied on evidence of assistance to the main tortfeasor to infer an agreement, and then attached the label "civil conspiracy" to the resultant amalgam. Sometimes, although not always, the inference has been factually justified; many tort defendants have both conspired with and substantially assisted each other. But we find it important to keep the distinctions clearly in mind as we review the facts in this novel case to see if tort liability is warranted on either or both concerted action theories. For the distinctions can make a difference. There is a qualitative difference between proving an agreement to participate in a tortious line of conduct, and proving knowing action that substantially aids tortious conduct. In some situations, the trier of fact cannot reasonably infer an agreement from substantial assistance or encouragement. A court must then ensure that all the elements of the separate basis of aiding-abetting have been satisfied. For example, Rael v. Cadena, 93 N.M. 684, 604 P.2d 822 (1979) (liability for verbal encouragement at the scene of a battery), discussed below, involved aid that was unlikely to support a conclusion of agreement. Furthermore, it is difficult to conceive of how a conspiracy could establish vicarious liability where the primary wrong is negligence, but a secondary defendant could substantially aid negligent action. The theory of liability also affects who is liable for what. An aider-abettor is liable for damages caused by the main perpetrator, but that perpetrator, absent a finding of conspiracy, is not liable for the damages caused by the aider-abettor.

We now proceed to examine case law in the context of this legal framework. Since this is a diversity case, we look first to District of Columbia law.10 Because the District's precedent is somewhat sparse in the area of concerted torts, we then push on to case law in other jurisdictions, which, although certainly not binding, is illustrative of some key issues.

District law acknowledges the concept of civil conspiracy, and assigns it the elements we outlined in section II. A., supra--basically, an agreement to take part in an unlawful action or a lawful action in an unlawful manner, and an overt tortious act in furtherance of the agreement that causes injury. Early on, the tort of civil conspiracy was described as follows: "The essence of conspiracy is an agreement--together with an overt act--to do an unlawful act, or a lawful act in an unlawful manner." Cooper v. O'Connor, 99 F.2d 135, 142 (D.C.Cir.), cert. denied, 305 U.S. 643, 59 S.Ct. 146, 83 L.Ed. 414 (1938). Accord Edwards v. James Stewart & Co., 160 F.2d 935, 936-37 (D.C.Cir.1947); International Underwriters, Inc. v. Boyle, 365 A.2d 779, 784 (D.C.1976).11 Subsequent cases emphasize that agreement can only lead to liability if an act pursuant to it causes injury. See DeBobula v. Goss, 193 F.2d 35, 36 (D.C.Cir.1951); Blankenship v. Boyle, 329 F.Supp. 1089, 1099 (D.D.C.1971) ("gist of a civil conspiracy ... is not the agreement ..., but the civil wrong ... done pursuant to the agreement"), motion for stay denied, 447 F.2d 1280 (1971) (per curiam), supplemented, 337 F.Supp. 296 (1972). The requirement of an actionable injury may explain why "there is no recognized independent tort action for civil conspiracy in the District of Columbia." See Waldon v. Covington, 415 A.2d 1070, 1074 n. 14 (D.C.1980) (emphasis added) (citing Lamont v. Haig, 590 F.2d 1124, 1136 n. 73 (D.C.Cir.1978)). Since liability for civil conspiracy depends on performance of some underlying tortious act, the conspiracy is not independently actionable; rather, it is a means for establishing vicarious liability for the underlying tort.

The separate tort of aiding-abetting has not yet, to our knowledge, been recognized explicitly in the District, but the existence of the civil conspiracy action suggests a high probability that the legal rationale underlying aiding-abetting would also be accepted: The District law recognizes that a person's actions in support of a wrong may make him liable for the tortious injury (i.e., civil conspiracy is only a means through which a plaintiff can establish vicarious liability, not an independent wrong). An agreement to participate in a wrongful course of action suffices to create vicarious liability. It seems likely that the District's courts would also find vicarious liability for support of wrongful action through knowing substantial aid or encouragement.

A few cases in other jurisdictions have considered more carefully than the District of Columbia two issues at the heart of Halberstam's conspiracy case against Hamilton: what kind of evidence is sufficient to establish an agreement to pursue wrongful conduct, and the extent of liability for acts performed by a coconspirator.

Both issues arose in Davidson v. Simmons, 203 Neb. 804, 280 N.W.2d 645 (1979), which involved a civil conspiracy to commit burglary. The boy whose liability was in question was the driver of a getaway truck. One of his codefendants, who had been inside the burglarized building, had struck an investigating policeman on the head with a hammer. Even though the driver never entered the building and had beaten a hasty retreat from the scene when the police appeared, the court held he could be liable to the injured policeman on a conspiracy basis. The court observed that "[a] conspiracy need not be established by direct evidence ... but may, and generally must, be proved by a number of indefinite acts, conditions, and circumstances which vary according to the purpose to be accomplished." The court stated that the finding of a conspiracy would be justified if the defendants' acts revealed that they had "pursued the same object, although by different means, one performing one part and another another part." 280 N.W.2d at 648-49 (citation omitted). On the issue of extent of liability, the Davidson court asserted that once the conspiracy had been established, all parties to it would be liable for injuries from acts pursuant to and in furtherance of the common design, even if the parties had not actively participated or benefited by the particular acts resulting in injury. Id. 280 N.W.2d at 649 (citation omitted).

The Davidson court cited a similar case: Tabb v. Norred, 277 So.2d 223 (La.App.), cert. denied, 279 So.2d 694 (1973), in which two armed boys burglarized a school and one shot an investigating officer. The boy who did not shoot the deputy argued that he was not liable because he had not fired the injurious shot and had not assisted in or encouraged the shooting; indeed, the boy had been disarmed by police before the deputy was wounded. The boy further argued that even if there had been a conspiracy, it had not contemplated gunfire. The court held that both boys were liable for the shooting under a statute that established liability for assisting or encouraging a person in the commission of an unlawful act. Its analysis, however, merged the conspiracy and aiding-abetting theories. The court based a finding of conspiracy on the boys' joint action of breaking into the school with pistols. Since both boys carried pistols and fired them at the police to avoid apprehension, the court found that the wounding shot was not beyond the scope of the conspiracy.

One might reasonably conclude from these cases that where two or more persons jointly commit an onsite burglary, a court will infer that there has been a prior agreement to do so, and that a violent act is within the reasonable scope of such an agreement, particularly when both persons are armed.

Peterson v. Cruickshank, 144 Cal.App.2d 148, 300 P.2d 915 (1956), extends the reasoning about inferring agreement to a situation where, as in this case, the parties did not execute the tort together at the same time and place. In Peterson, the issue was "whether there [was] any substantial evidence to support the finding that appellant conspired with his two co-defendants to falsely imprison [appellant's consort]," id. 300 P.2d at 925, in a sanitarium where she received shock treatments. The appellant protested that all he had done was pay his consort's bills; he had neither directed the doctor at the sanitarium to imprison her or administer shock treatments. The court observed that, absent a confession, an agreement between conspirators must generally be inferred from circumstantial evidence revealing a common intent; it found a number of circumstances that permitted the inference that appellant had reached an understanding with his codefendants about the restraint and treatment of his consort. First, appellant's past stormy personal relationship with the woman had provided him a motive to have her restrained. Second, there was evidence of a conversation within a few days of the confinement between appellant and a codefendant doctor at the sanitarium. In this conversation, appellant and the doctor had discussed appellant's falling out with the woman, the history of appellant's relationship with her, and appellant's willingness to pay all bills for her "treatment." After the talk, the doctor had refused to let the woman's sister take her home; he also secured, under suspicious circumstances and over her sister's objection, the woman's "consent" to shock treatments. Furthermore, before the woman left the sanitarium, appellant sent an attorney to induce her to sign a release of "all claims." The court found such evidence sufficient to sustain the finding of a conspiracy between appellant and the doctor to imprison the woman against her will.

Davidson, Tabb, and Peterson provide some insights into the amount and kind of evidence necessary to establish prior agreement as well as into the extent of liability for a coconspirator's acts. Davidson and Peterson in particular recognize that since in most cases the court will have to infer a conspiracy from indirect evidence, it must initially look to see if the alleged joint tortfeasors are pursuing the same goal--although performing different functions--and are in contact with one another. The circumstances of each case dictate what other specific evidence may be useful in inferring agreement. The easiest situation in which to draw the inference of agreement is where the parties are on the scene together at the same time performing acts in support of one another. This description approximates Tabb: two armed persons travel together to a building, both break in, and both shoot when confronted by police. In Davidson, the defendants were performing different but connected acts, relatively close in time and location: driving the getaway vehicle and breaking in. The performance of different acts at different times in different places, as in Peterson, requires a more extensive set of inferences to link the actors together. But such inferences are still sustainable in the proper factual setting. Additionally, the length of time two parties work closely together may also strengthen the likelihood that they are engaged in a common pursuit. Mutually supportive activity by parties in contact with one another over a long period suggests a common plan.

In sum, we expect that the relationships between the actors and between the actions (e.g., the proximity in time and place of the acts, and the duration of the actors' joint activity) are relevant in inferring an agreement in a civil conspiracy action. There may well be other significant factors in individual cases.

As to the extent of liability, once the conspiracy has been formed, all its members are liable for injuries caused by acts pursuant to or in furtherance of the conspiracy. A conspirator need not participate actively in or benefit from the wrongful action in order to be found liable. He need not even have planned or known about the injurious action, as in the case of the getaway driver in Davidson, so long as the purpose of the tortious action was to advance the overall object of the conspiracy.

Our second line of cases focuses on the aiding-abetting theory of liability--specifically on what constitutes knowing substantial assistance or encouragement and on the extent to which an aider-abettor is liable for injuries caused by the principal tortfeasor. Calling attention again to the distinction between aiding-abetting and civil conspiracy, we find both cases that are "pure" aiding-abetting and ones that courts could probably also have found to be civil conspiracies.

An example of the purer strain, Rael v. Cadena, 93 N.M. 684, 604 P.2d 822 (1979), involved a person who had given verbal encouragement ("Kill him!" and "Hit him more!") to an assailant. The defendant had not physically assisted in the battery. The court explained that liability did not require a finding of action in concert, nor even that the injury had directly resulted from the encouragement. Instead, it found, citing Restatement Sec. 876(b), that the fact of encouragement was enough to create joint liability for the battery. Mere presence at the scene, it noted, would not be sufficient for liability.

Suggestive words may also be enough to create joint liability when they plant the seeds of action and are spoken by a person in an apparent position of authority. In Cobb v. Indian Springs, Inc., 258 Ark. 9, 522 S.W.2d 383 (1975), a security guard allegedly urged a younger motorist with a new car to "run [the car] back up here and see what it will do." 522 S.W.2d at 387. The driver then struck the plaintiff while trying to avoid a pedestrian during his high-speed "test run." The court held, relying on Restatement Sec. 876(b), that a jury could have found the guard's encouragement substantial because he had first proposed the trial drive and because his position of authority gave his suggestion extra weight. On the issue of extent of liability, the Cobb court found that the guard could have foreseen an appreciable risk of harm to others at the time of encouragement.

Vicarious liability can of course be based on acts of assistance as well as words of encouragement. And the contributing activity itself need not be so obviously nefarious as cheering a beating or prodding someone to drive recklessly. Keel v. Hainline, 331 P.2d 397 (Okl.1958), involved students throwing erasers at one another in a classroom. One eraser struck the plaintiff, a nonparticipant in the "horse play"; her eye glasses shattered and she lost the use of an eye. The court found that a student who had only aided the throwers by retrieving and handing erasers to them was still liable for the injury, because he had substantially encouraged the wrongful activity that resulted in the injury. It did not matter that the defendant may not even have given any particular aid to the boy who threw the eraser that hit the plaintiff.

Rael, Cobb, and Keel dealt with direct encouragement by word or deed at the scene of the tort. But the aiding-abetting action may also be more distant in time and location and still be substantial enough to create liability. In Russell v. Marboro Books, 18 Misc.2d 166, 183 N.Y.S.2d 8 (N.Y.Sup.Ct.1959), the court held that a book company could be liable as a contributing tortfeasor because it had sold a model's picture to a company with the knowledge the company would (as it did) alter and use the picture to defame the model. Relying partially on Restatement Sec. 876(b), the court reasoned that acquisition of the photograph was an indispensable prerequisite to the libel. The sale, along with knowledge of the buyer's intent to alter and publish the picture, could constitute "substantial assistance" for joint tort liability.

Russell, however, falls in the category of cases that might have been alternatively analyzed as civil conspiracies. Indeed, to the extent that the acts of aiding-abetting occur further from the actual scene of the tortious injury, the cases become more readily treatable as either conspiracies or substantial assistance cases. It is not always clear why courts take one theory over the other; sometimes they rely on both. As we said earlier, this may be permissible, but for reviewing purposes it is important to keep in mind the conceptual differences between the two torts, so that we can check that the necessary proof for one or both has been supplied, rather than simply a little of each.

American Family Mutual Insurance Co. v. Grim, 201 Kan. 340, 440 P.2d 621 (1968), is a good example of judicial merger of the theories. Grim involved the liability of a thirteen-year-old boy to an insurer of a church damaged by fire. The boy had broken into the church with his companions at night in search of soft drinks in the kitchen. Two other boys failed to extinguish torches they had used to light their way in the church attic. As a result, the church was damaged by an extensive fire. The boy whose liability was in question neither entered the attic nor even knew about the torches. He was not near the church when the fire appeared. Nevertheless, the court found the boy liable for the fire damage. In reaching this conclusion, the Grim court appeared to rely in part on a conspiracy-type analysis, reasoning that the boy was liable, despite his lack of involvement with the torches, because "the torches were used in the four boys' attempt to carry out their original unlawful plan." Id. 440 P.2d at 625. But the Grim court also drew upon aiding-abetting concepts. It observed that the defendant was more than an innocent bystander after the boys had entered the church: "[T]here was evidence from which it could be inferred that [the boy] actively participated and lent encouragement and cooperation to the successful accomplishment of their over-all mission." Id. 440 P.2d at 626. In sum, the Grim court was invoking both civil conspiracy and aiding-abetting theories on the ground there was both an agreement to break in to get soft drinks and substantial aid through actual participation in the break-in.

The facts of Grim--a break-in to pilfer soft drinks by four boys, two of whom jerry-rigged torches that caused substantial fire damage--also raise an interesting question of the permissible extent of liability: When is a defendant liable for injuries caused by the acts of the person assisted when the acts were not specifically contemplated by the defendant at the time he offered aid? The Grim court, drawing on the commentary to Restatement Sec. 876(b), pointed out that the principle to apply in assigning liability under the aiding-abetting theory was: "[A] person who encourages another to commit a tortious act may also be responsible for other foreseeable acts done by such other person in connection with the intended act." Id. 440 P.2d at 626. It then referred to a germane illustration carried forward in the present Restatement:

A and B conspire to burglarize C's safe. B, who is the active burglar, after entering the house and without A's knowledge of his intention to do so, burns the house in order to conceal the burglary. A is subject to liability to C, not only for the conversion of the contents of the safe but also for the destruction of the house.

Restatement Sec. 876, comment d, illustration 10. The example, however, obviously involves a conspiracy, not mere aiding-abetting. Looking at the facts of its own case, the Grim court noted that "the need for adequate lighting could reasonably be anticipated [and] torches served that purpose." 440 P.2d at 626. Thus, the boy who had not used a torch, nor even expected one to be lighted, could be liable for the damage caused by the torches because their employment was foreseeable.12

Before leaving the aiding-abetting line of cases, it is helpful to review a case in which a court found that there was not enough evidence of assistance to support liability. Duke v. Feldman, 245 Md. 454, 226 A.2d 345 (1967), involved an allegation that a woman was liable for civil assault because she had aided and assisted her husband, who had struck the plaintiff. To establish a claim against the wife, the plaintiff would have had to present "evidence that she assisted, supported, or supplemented her husband's action or that she instigated, advised, or encouraged the commission of the tort." Id. at 348, 226 A.2d 345 (citation omitted). Evidence that she was merely present at or took pleasure in the assault and battery would not be enough to create liability. The plaintiff's evidence of her involvement--consisting of the defendant's awareness of her husband's previous threats to plaintiff; her contemporaneous request to her husband to get their downpayment back from the plaintiff; her observation of the incident; and driving her husband away--was found insufficient to go to the jury. The Duke court observed that, based on this evidence, it would have been pure speculation to hold that the defendant had been aiding her husband. Id. at 348, 226 A.2d 345.

Summing up our review of the aiding-abetting cases, it is obvious that many variables entered into the equation on how much aid is "substantial aid." Generally, the cases support the five factors identified in the Restatement: the nature of the act encouraged; the amount [and kind] of assistance given; the defendant's absence or presence at the time of the tort; his relation to the tortious actor; and the defendant's state of mind. See Restatement Sec. 876(b), comment d.

For example, the nature of the act involved dictates what aid might matter, i.e., be substantial. In Rael, the beating case, the defendant's war cry for more blood may well have contributed to the assaulter's hysteria, which was fueling his physical acts of violence. Obviously verbal support would have been of lesser import in Russell, the defamed model case, where the key assistance rendered was access to the photo.13

The amount [and kind ] of assistance given the wrongdoer is also a significant factor in the cases. For example, the court in Cobb stressed the security guard's major part in prompting the tort--his suggestion sparked the negligent action. Similarly, the sale of the model's photo in Russell was integral to the wrongful alteration and advertisement. Presence at the time of the tort, the third factor, applies to all the aider-abettor cases discussed above except Russell. On the other hand, the court in Duke stressed that an assailant's wife was not jointly liable, despite her presence, because she offered no real assistance.

The fourth factor, the defendant's relation to the tortfeasor, was significant in Cobb, where the court emphasized that the guard's position of authority lent greater force to his power of suggestion. The Keel court also seemed to believe that the students' group action, their creation of a free-for-all, was both dangerous and ultimately injurious; thus, it found even a minimally-involved participant liable. This focus on group activity giving rise to joint liability for the wrongs of the group also appears in Grim (the church break-in and fire). Of course, the intimacy of the relationship between the wrongdoer and the defendant may also lead a court to be wary of inferring assistance warranting joint liability from supportive activities. Thus, in Duke the court seemed especially vigilant in evaluating evidence of the wife's assistance to her husband (who assaulted someone while she watched), so as not to infuse the normal activities of a spouse with the aura of a concerted tort.

Fifth, evidence as to the state of mind of the defendant may also be relevant to evaluating liability. In Rael, the defendant's abusive cheering of the battery showed he was one in spirit with the assaulter. But once again, the Duke court (wife on the scene of her husband's assault) refused to find evidence of liability, despite the fact that the wife had urged her husband to get the payment and had watched silently while the assault took place.

Finally, we would also rate a sixth factor--duration of the assistance provided --as important. The length of time an alleged aider-abettor has been involved with a tortfeasor almost certainly affects the quality and extent of their relationship and probably influences the amount of aid provided as well; additionally, it may afford evidence of the defendant's state of mind.

As for the second issue in aiding-abetting, the extent of liability, the test from Cobb and Grim appears to be that a person who assists a tortious act may be liable for other reasonably foreseeable acts done in connection with it. While this language is slightly different from that found in civil conspiracy cases--where a conspirator is liable for acts pursuant to, in furtherance of, or within the scope of the conspiracy--we are not sure that it is a distinction that makes a practical difference. Foreseeability is surely an elusive concept and does not lend itself to abstract line-drawing. The court in Grim, citing the Restatement, apparently did not think conspiracy and aiding-abetting warranted different tests for extent of liability. We need not look further into this matter here, however, because we find below that Hamilton is liable for Halberstam's death under the language of both tests of extent of liability.

We note, finally, that the concept of tortious aiding-abetting has turned up frequently in the evaluation of secondary liability for securities law violations, principally in the area of fraud. In the main, the securities cases provide support for the distinctions we have drawn between the elements of aiding-abetting and civil conspiracy.14 Some even rely on the Restatement, as we have, to help distinguish between the two concepts. See, e.g., Pharo v. Smith, 621 F.2d 656, 669 (5th Cir.1980). They also underscore why line-drawing between the two theories matters. For example, the court in Epprecht v. Delaware Valley Machinery, Inc., 407 F.Supp. 315, 320 (E.D.Pa.1976), noted that liability may be based on a more attenuated relation with the principal violation in a conspiracy than in aiding-abetting. In a sense, the agreement in a conspiracy may substitute for the "substantiality" of an aider-abettor's assistance in carrying out the violation, thereby allowing greater temporal or physical distance between the conspirator and the wrongful act. The securities cases have also employed the Restatement list of evaluative factors on "substantial assistance." See, e.g., Monsen v. Consolidated Dressed Beef Company, 579 F.2d 793, 800 (3d Cir.), cert. denied, 439 U.S. 930, 99 S.Ct. 318, 58 L.Ed.2d 323 (1978); Landy v. Federal Deposit Insurance Corp., 486 F.2d 139, 162-63 (3d Cir.1973), cert. denied, 416 U.S. 960, 94 S.Ct. 1979, 40 L.Ed.2d 312 (1974).15

Our first step in applying the law to the facts of this case is to review the district court's findings of fact under the "clearly erroneous" standard. Fed.R.Civ.P. 52(a). This standard applies to the inferences drawn from findings of fact as well as to the findings themselves.16 The test places considerable limits on our discretion.

The basic facts in the case are undisputed and, as set out in Part I of the opinion, grounded almost entirely in Hamilton's testimony. It is the inferences the trial judge drew from those facts that are in contention on appeal. And it is the district court's inferences that are essential to establishing the elements of civil conspiracy and aiding-abetting in this case. First, the district court found that Hamilton "knew full well the purpose of [Welch's] evening forays and the means" he used to acquire their wealth. See District Court Opinion at 5, 6. Second, the district court inferred an agreement--that "[she] was a willing partner in his criminal activities." See id. Third, the district court pointed to various acts by Hamilton (e.g., typing transmittal letters for the ingot sales, handling the payments and accounts, maintaining all financial transactions solely in her name), see id. at 4, and concluded that they were performed knowingly to assist Welch in his illicit trade: "Disposing of the loot was the principal business in which Welch and Hamilton engaged while at home. Hamilton worked as secretary and recordkeeper of their transactions ..." Id. at 5. See also id. at 6 (in its conclusions of law, the court noted Hamilton "knowingly and willingly assisted in Welch's burglary enterprise").

Based upon the record before us, we do not find these inferences to be impermissible. The facts lend them substantial support. The district court also emphasized that its conclusions were based in part on "the demeanor and behavior of Hamilton under oath." Id. at 5.

As to the inference of Hamilton's knowledge of Welch's criminal doings, it defies credulity that Hamilton did not know that something illegal was afoot. Welch's pattern of unaccompanied evening jaunts over five years, his boxes of booty, the smelting of gold and silver, the sudden influx of great wealth, the filtering of all transactions through Hamilton except payouts for goods, Hamilton's collusive and unsubstantiated treatment of income and deductions on her tax forms, even her protestations at trial that she knew absolutely nothing about Welch's wrongdoing--combine to make the district court's inference that she knew he was engaged in illegal activities acceptable, to say the least.

Similarly, the district court's finding of an agreement between Welch and Hamilton to execute a criminal enterprise involving stolen goods was not "clearly erroneous." As we discussed above, courts have to infer an agreement from indirect evidence in most civil conspiracy cases. The circumstances of the wrongdoing generally dictate what evidence is relevant or available in deciding whether an agreement exists. Factors like the relationship between the parties' acts, the time and place of their execution, and the duration of the joint activity influence the determination. In this case, Hamilton and Welch did not commit burglaries together but their activities were symbiotic. They were pursuing the same object by different but related means. Their home became the storage and processing base for Welch's criminal activities; they thus performed some of their different parts of the illegal operation together at the same location. The long-running nature of the scheme is also crucial to the inference of agreement--Hamilton's knowledge and aid over five years makes some kind of accord extremely likely--perhaps only a tacit accord, but that is enough. Furthermore, while Hamilton's extensive participation in the profits of the illegal venture might not by itself prove an agreement, her unquestioning accession of wealth during this period is certainly consistent with such an agreement. Totaling all this evidence up, the district court's conclusion that Hamilton and Welch reached an understanding about their illegal enterprise withstands attack.

Finally, the district court's inference of knowing assistance also stands under the "clearly erroneous" test. Hamilton's invaluable service to the enterprise as banker, bookkeeper, recordkeeper, and secretary is substantiated by her own testimony. She performed these services in an unusual way under unusual circumstances for a long period of time and thereby helped launder the loot and divert attention from Welch. Given all this, we will not upset the court's inference that she knew she was assisting Welch's wrongful acts. As the Supreme Court noted, "[a] finding is 'clearly erroneous' [only] when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed." United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 541, 92 L.Ed. 746 (1948).

The district court relied on the same three factual inferences to conclude that Hamilton was liable as a coconspirator. District Court Opinion at 6. We agree. To summarize our earlier discussion, in the District of Columbia a conspiracy requires: an agreement to do an unlawful act or a lawful act in an unlawful manner; an overt act in furtherance of the agreement by someone participating in it; and injury caused by the act. We have upheld the district court's finding that Hamilton and Welch agreed to undertake an illegal enterprise to acquire stolen property. The only remaining issue, then, is whether Welch's killing of Halberstam during a burglary was an overt act in furtherance of the agreement. We believe it was. As noted above, a conspirator can be liable even if he neither planned nor knew about the particular overt act that caused injury, so long as the purpose of the act was to advance the overall object of the conspiracy. Welch was trying to further the conspiracy by escaping after an attempted burglary, and he killed Halberstam in his attempt to do so. The use of violence to escape apprehension was certainly not outside the scope of a conspiracy to obtain stolen goods through regular nighttime forays and then to dispose of them. Cf. Davidson v. Simmons, 203 Neb. 804, 280 N.W.2d 645 (1979) (driver of the getaway truck found liable as a coconspirator for an unplanned injury to a policeman inflicted by his inside partner, who was caught in the course of the burglary). In sum, the district court's findings that Hamilton agreed to participate in an unlawful course of action and that Welch's murder of Halberstam was a reasonably foreseeable consequence of the scheme are a sufficient basis for imposing tort liability on Hamilton according to the law on civil conspiracy.

The district court also concluded that Hamilton was liable as a "joint venturer." The elements to which the court referred--that Hamilton knew of Welch's illegal activity and assisted in it--suggest to us that the court basically relied on the theory that we have labeled aiding-abetting. We have summarized its elements as follows: (1) the party the defendant aids must perform a wrongful act that causes an injury; (2) the defendant must be generally aware of his role as part of an overall illegal or tortious activity at the time he provides the assistance; and (3) the defendant must knowingly and substantially assist the principal violation.

Welch fulfilled the first of these three elements by killing Halberstam during the course of a burglary. The district court's conclusions that Hamilton knew about and acted to support Welch's illicit enterprise establish that Hamilton had a general awareness of her role in a continuing criminal enterprise. The second element is thus satisfied. Finally, the district court also justifiably inferred that Hamilton assisted Welch with knowledge that he had engaged in illegal acquisition of goods. The only remaining issue, then, is whether her assistance was "substantial."

Applying the Restatement's five factors, we look first at the nature of the act assisted, here a long-running burglary enterprise, heavily dependent on aid in transforming large quantities of stolen goods into "legitimate" wealth. Hamilton's assistance was indisputably important to this laundering function; she gave not only her time and talents but also her name to accomplish that objective, through having checks made out to her and falsifying income tax returns. Although her own acts were neutral standing alone, they must be evaluated in the context of the enterprise they aided, i.e., a five-year-long burglary campaign against private homes. Second, although the amount of assistance Hamilton gave Welch may not have been overwhelming as to any given burglary in the five-year life of this criminal operation, it added up over time to an essential part of the pattern.

Third, Hamilton was admittedly not present at the time of the murder or even at the time of any burglary. But as we noted above, the success of the tortious enterprise clearly required expeditious and unsuspicious disposal of the goods, and Hamilton's role in that side of the business was substantial.

Fourth, the significance of Hamilton's relation to the tortious actor requires a careful balancing. We are understandably wary of finding a housemate civilly liable on the basis of normal spousal support activities. Even though Hamilton's assistance surely transcended performing household chores for Welch, we must be cautious not to overemphasize the relationship. Hence, we accord it a low priority in our calculus.

On the other hand, the fifth factor, the defendant's state of mind assumes a special importance in this case. If, as the district court found, Hamilton's assistance was knowing, then it evidences a deliberate long-term intention to participate in an ongoing illicit enterprise. Hamilton's continuous participation reflected her intent and desire to make the venture succeed; it was no passing fancy or impetuous act. Finally, the duration of the assistance has strongly influenced our weighing of Hamilton's assistance. It affected our sense of how Hamilton perceived her role and of the value of her assistance to Welch. In sum, we find that Hamilton's assistance was indeed substantial enough to justify liability on an aider-abettor theory.

On the scope of her liability, we agree with the district court that Hamilton's assistance to Welch's illegal enterprise should make her liable for Welch's killing of Halberstam. We noted above that under a civil conspiracy theory, it was within the scope and in furtherance of their agreement to conduct the illicit burglary enterprise. Similarly, under an aiding-abetting theory, it was a natural and foreseeable consequence of the activity Hamilton helped Welch to undertake. It was not necessary that Hamilton knew specifically that Welch was committing burglaries. Rather, when she assisted him, it was enough that she knew he was involved in some type of personal property crime at night--whether as a fence, burglar, or armed robber made no difference--because violence and killing is a foreseeable risk in any of these enterprises. Cf. American Family Mutual Insurance Co. v. Grim, 201 Kan. 340, 440 P.2d 621 (1968) (church break-in foreseeably leads to fire caused by unextinguished torches); Restatement Sec. 876(b), comment d, illustration 10 (the action of a burglary conspirator who burns a home to conceal the crime is foreseeable).

We have lingered long on the elements of traditional tort theory that permit holding a nonparticipant in a burglary that led to murder civilly responsible for the economic consequences of so terrible an injury. Our effort to distinguish the elements and proof of civil conspiracy and aiding-abetting may appear formalistic, but it is motivated by our desire to move cautiously in cases like this one. Our ultimate purpose is to identify those characteristics that make the application of vicarious liability appropriate. We recognize that the elements of either theory are not perfect guides in this search. We expect that they will not be accepted as immutable components but that they will be adapted as new cases test their usefulness in evaluating vicarious liability.

Tort law is not, at this juncture, sufficiently well developed or refined to provide immediate answers to all the serious questions of legal responsibility and corrective justice. It has to be worked over to produce answers to questions raised by cases such as this. Precedent, except in the securities area, is largely confined to isolated acts of adolescents in rural society. Yet the implications of tort law in this area as a supplement to the criminal justice process and possibly as a deterrent to criminal activity cannot be casually dismissed. We have seen the evolution of tort theory to meet twentieth century phenomena in areas such as product liability; there is no reason to believe it cannot also be adapted to new uses in circumstances of the sort presented here. This case is obviously only a beginning probe into tort theories as they apply to newly emerging notions of economic justice for victims of crime. For present purposes, we are satisfied that the district court's factual findings and inferences fit into existing concepts of civil liability for concerted tortious actions through conspiracy and aiding-abetting. The judgment of the district court imposing civil liability on Hamilton for Halberstam's death is therefore

Affirmed.