Alexander Grant & Co., an accounting firm, appeals the trial court's refusal to overturn an $11 million jury verdict for negligence, fraud, and breach of contract in the preparation of auditing statements for Auto-Train, a corporation that later declared bankruptcy. We reverse, finding no evidence of a causal relation between the accountant's failures and the client's damages.

* * * * * *

Auto-Train was a promising idea that went sour. During ten years of operation, beginning in 1971, the company carried passengers and their cars between Lorton, Virginia (near Washington, D.C.) and Sanford, Florida (near Orlando). Those who travelled by Auto-Train saved either the trouble of driving to Florida or the expense of flying and renting a car on arrival. This attraction apparently worked well enough until 1976, the company's first year in the red. It never recovered, filing for bankruptcy in September 1980. Eight months later the trains stopped rolling. Liabilities were in the neighborhood of $25 million and assets "minimal."

Grant served as Auto-Train's independent outside auditor. The bankrupt's trustee claimed various deficiencies in Grant's annual audits of company financial statements for the years 1976 to 1979, and in a (pre-bankruptcy) 1980 opinion on that of a subsidiary. The most significant was Grant's failure to make a specific note of Auto-Train's failure to pass on to the IRS payroll taxes withheld from employees; instead of doing so on a timely basis, Auto-Train used these funds to meet its own cash needs. The trustee also claimed that Grant grossly overstated the value of Auto-Train's rolling-stock and spare parts inventory; the correct figures would have made Auto-Train's disastrous financial statements appear still more calamitous. The trustee contrasts the $1.2 million in proceeds for liquidation of rolling stock with the $18 million figure at which it was carried on the financial statements.

As to both defects, the trustee argued that proper disclosure by Grant would have prompted an alert among the company's directors at a time when corrective action was still possible. At the very least, says the trustee, the company would have filed for bankruptcy protection earlier than it did, thus saving millions in liabilities needlessly accumulated in the doomed effort to keep Auto-Train running. At trial the trustee asked for $25 million in damages (roughly the bankrupt's entire negative net worth) on theories of fraud, negligence, and breach of contract. By special verdict, the jury gave him $11 million, finding Grant liable on all three theories, but only for the 1978 and 1979 audits. The trial court denied Grant's motion for judgment notwithstanding the verdict, and Grant appeals.

* * * * * *

Although the parties feud a good deal over concepts of legal causation, this is not a case that turns on refinements of those concepts. In the trustee's two theories of harm--i.e., of rescue obstructed or liquidation deferred--the principle is that accurate audits would have prompted company action to cure or mitigate the company's basic economic problems. Under each theory, Grant's omissions are characterized as a discrete link in a chain of causation. Thus the facts do not fall within the limited class of cases where courts relieve the plaintiff of the minimal duty of showing "but for" causation--cases where multiple independent causes (e.g., a fire caused by negligence and one of unknown origin), each capable of bringing about a loss, operate coincidentally. See W. Page Keeton et al., Prosser and Keeton on the Law of Torts 266-67 (5th ed. 1984). Accordingly, the trial court here was required to grant the defendant's motion for judgment n.o.v. if no reasonable person could find it more probable than not that correct accountancy would have averted either of the harms alleged. See, e.g., Waugh v. Suburban Club Ginger Ale Co., 167 F.2d 758 (D.C.Cir.1948); see also Cotton v. Buckeye Gas Products Co., 840 F.2d 935, 937 (D.C.Cir.1988) (reviewing federal and District of Columbia standards for grant of directed verdict or judgment n.o.v.). This sort of causal link is necessary regardless of whether the defendant failed some test of professional care, and it applies to all three theories of recovery.

It is undisputed, of course, that Grant was not responsible for the underlying cause of Auto-Train's failure. As a result of recurrent derailments caused by braking failures, Auto-Train had to shorten the length of its trains and increase its fares at the same time that the airlines were discounting theirs. See Joint Appendix ("J.A.") 186-87 ("Management's Discussion and Analysis of Summary of Operations" for 1979). Bad luck both at the company itself and in its market, and perhaps bad management, brought about the losses. Grant's place on the causal chain consists simply of its alleged failure to raise the alarm.

But Grant had raised the alarm. Its audits hardly painted a rosy picture of Auto-Train's financial condition. For the years for which Grant was found liable, 1978 and 1979, they showed losses of $4.4 million and $2.8 million, respectively--no cause for celebration or even complacence. For both of those years, moreover, Grant issued what are known in accounting shorthand as "going concern" qualifications, raising the possibility that the subject of the audit will not keep "going" and warning that in the event of a shutdown the balance-sheet asset valuations would not be fully realized. See J.A. 114, 128 (qualifications to Grant's 1978 financial statement); J.A. 225, 234 (for 1979); see generally Copy-Data Systems, Inc. v. Toshiba America, Inc., 755 F.2d 293, 299 & n. 2 (2nd Cir.1985); Devaney v. Chester, 1989 WL 52375, 1989 U.S.Dist. LEXIS 4986, slip op. at 4 (S.D.N.Y.1989). Issuing a going concern opinion may not insulate an accounting firm from liability, cf. Bradford-White Corp. v. Ernst & Whinney, 1987 WL 16316, 1987 U.S.Dist. LEXIS 7879, slip op. at 36-37 (E.D.Pa.1987), aff'd in part, rev'd in part, 872 F.2d 1153 (3rd Cir.1989), but it must cut strongly in its favor. See In re North American Acceptance Corp. Securities Cases, 513 F. Supp. 608, 636 n. 15 (N.D.Ga.1981) (calling "going concern" qualification "about the most conspicuous 'red flag' that an auditor can wave"). As Auto-Train's condition was so grimly depicted, the trustee is reduced to claiming that the alarm wasn't raised loudly enough, that another decibel or two would have made all the difference.

The trustee nonetheless advances a scenario, hinging on the role of Auto-Train's outside (i.e., non-management) directors, in which Grant appears as the key villain. He contends that these directors, who filled four of seven spots on the company's board (three of six after a vacancy was created by resignation in 1979), did not know of the tax or valuation problems, and that if they had known, they would have either taken necessary corrective action to restore the company's financial health or put it immediately into receivership, thus cutting off the accrual of its eventual $25 million in net liabilities.

For this the trustee relies on the litigative equivalent of a sound bite in the midst of a ten-week trial. Only one outside director, George L. Green, gave any indication at trial that awareness of the payroll tax problem might have spurred him on to decisive action, or indeed to any action at all. If Grant's audit had specified the delinquency, Green testified on direct examination,

A: As a director, I could have insisted that management reduce expenses, reduce overhead, eliminate things that could have waited to accumulate payments, monies, so that we could liquidate that obligation on the payroll taxes.... If that did not succeed, I would strongly recommend, because of the fact that that monies [sic] were being used to finance the company and we could not raise the capital, then I think my position would have been that--

The Court: Well, what action could you have taken is the question.

A: I could have recommended filing for bankruptcy.

J.A. 1113-14.

We assume the jury could reasonably have credited this statement. But compare Devaney, slip op. at 14 ("[v]ague, self-serving, speculative testimony concerning what a party would have done under different circumstances is generally not admissible"); Matter of Hawaii Corp., 567 F. Supp. 609, 627 (D.Hawaii 1983) (in bench trial court discounts as "unreliable" testimony of board members who approved a reorganization that later bombed and who then claimed that more accurate accountant's statements would have turned their votes). Even so, plaintiff has failed to establish the minimum necessary probability that Grant's omissions materially prolonged Auto-Train's decline. See Chase v. Gilbert, 499 A.2d 1203, 1212 (D.C.1985).

No one seriously questions that the inside directors, though well enough aware of the problem (perhaps even its source), stood by silently and passively. As for the outside directors other than Green, none even so much as speculated as to how he might have responded to the tax revelation. Indeed, only one of them, Jerald Jarrard, testified with any certainty that he was ignorant of the tax delinquency despite Grant's omission, J.A. 1124, while conceding that he "didn't have to read the annual report" to know that the company was "in pretty sad shape." J.A. 1188. The trustee's case is thus reduced to the assertion of one director (Green) that but for Grant's omission he would have insisted on things one would expect a director of a desperate company to have been pushing for in any case (e.g., reduce expenses) or, failing that, "have recommended" filing for bankruptcy.

The trial judge recognized the necessarily speculative character of Green's own testimony, and indeed for that reason insisted that it be framed in the highly contingent form of saying that he "could " do this or that, rather than that he would have. See J.A. 1104-13. But the weakness of the testimony goes further. Green's role as a single director limited his ability to effect a rescue or pull the plug. Action would have required the support of the three other outside board members (assuming support from their management colleagues unlikely, given their knowledge, passivity, and possible connivance). There is no evidence that Green could have persuaded them that immediate bankruptcy was the right response to the tax delinquency. In Day v. Avery, 548 F.2d 1018 (D.C.Cir.1976), we discounted a somewhat comparable claim. A law partner claimed injury from the firm executive committee's falsely asserting that no partner would be worse off as a result of a merger; in fact the merger triggered a coup removing Day as head of a branch office. He claimed that in not opposing the merger he had relied on the committee's promise. The court found that reliance immaterial to the final decision to merge, as "appellant's vote was not necessary to effect the merger nor sufficient to defeat it," id. at 1028, and upheld summary judgment for the partnership.

Of course it might seem easier for Green to have rallied the outside directors around a company rescue than for Day to have rallied his partners to his personal cause. But in fact we know here what actually happened when the tax specifics did become known: nothing. In August 1980 an article in The Washington Post revealed the unpaid tax liability, and an audit committee meeting, with Green and one other outside director in attendance, broached the subject. The directors did not swing into immediate action, but rather were quite content with an assurance that the IRS was not about to seize assets to satisfy the deficiency--and with, at the same time, renewing Grant's contract as auditor. See J.A. 347 (audit committee minutes); J.A. 1098-99 (Green testimony); Reply Brief of Appellant 11 n. 15. The company's eventual bankruptcy was in fact prompted not by the tax problem, but by its failure to secure continued financing. See J.A. 1205 (testimony of Auto-Train CEO Eugene Garfield).

The trustee's case is even weaker with respect to the alleged overvaluation of the company's assets. On that score the going concern qualifications were explicit, warning that a "major portion of the assets reflected in the accompanying balance sheets is dependent upon the Company's ability to meet its financing requirements, to maintain present financing, and the success of its future operations." J.A. 128 (for 1978); J.A. 234 (for 1979) (emphasis added). It also seems evident that the outside directors had some idea of the state of the idle equipment; at a June 1979 board meeting, one quizzed management as to the condition of the company's rolling stock, expressing concern over its "appearance and maintenance." J.A. 681. See Devaney, supra, slip op. at 12-13 (no causation where plaintiff had knowledge of facts it claimed another should have provided). Nothing appears on the other side of the balance. No one, not even Green, professed ignorance of potential problems with the asset valuations (assuming Grant to have been in some way derelict with respect to them), much less that enlightenment would have ushered in a new regime.

* * * * * *

We are of course hesitant to nullify a jury's findings of fact. See Prosser and Keeton on Torts 264-65. But the jury's primacy as factfinder cannot exempt it from the bounds of reasonableness. A defendant who is clearly not responsible for a harm should not have to pay for it. Cf. Bastian v. Petren Resources Corp., 892 F.2d 680, 685 (7th Cir.1990) ("[d]efrauders are a bad lot and should be punished, but Rule 10b-5 does not make them insurers against national economic calamities"); DiLeo v. Ernst & Young, 901 F.2d 624, 628-29 (7th Cir.1990) (unwarranted extension of accountant liability would reduce investor information as increase in cost of accountancy services would diminish quantity purchased). The court submits to the jury only those issues on which the party with the burden of proof offers "significantly probative" evidence. See Richardson v. Richardson-Merrell, Inc., 857 F.2d 823, 829 (D.C.Cir.1988). The trustee's case fails at this threshold. This judicial screening undoubtedly involves the court in some gauging of probabilities, but only at the extremes. See Wex S. Malone, Ruminations on Cause-In-Fact, 9 Stan.L.Rev. 60, 68-72 (1956). Performance of the task is an essential backstop to judicial definition of the law; if courts' endorsement of verdicts is automatic, the law, however sound, can simply drift off into irrelevance as juries do their will.

Finding no significant evidentiary foundation for the jury's finding of liability, we reverse.

So ordered.