This is a Federal Tort Claims suit to recover damages allegedly caused by the government’s negligence. There is no dispute as to the relevant underlying facts. Only the district court’s findings and conclusions with reference to causation are questioned.

During the latter half of 1965 (August to December) Jessee Swanner, a convicted felon, was employed as an informer for the Alcohol Tax Unit of Internal Revenue (ATU) in an undercover investigation of an illicit whiskey operation in Giles County, Tennessee. The acknowledged leader of this whiskey ring was a character named Ed McGlocklin who had a history and reputation in Giles County as a man of violence.

One of plaintiffs’ witnesses testified that during a conversation he had with McGlocklin in January 1966 the latter stated “that Jessee Swanner was the one that turned him up” and that if he could find Swanner “he would never appear in court to testify against him”. This witness also testified that in response to McGlocklin’s inquiry he told him that he heard Swanner was living in Montgomery, Alabama. Swanner learned indirectly of this threat on April 18, 1966, and promptly alerted the appropriate representatives of ATU. The ATU assured him that nothing would happen as long as he stayed in Alabama but cautioned him against going into Tennessee. On the evening of April 22, 1966, about midnight, three days before Swanner was scheduled to testify against the members of the whiskey ring before a federal grand jury in Giles County, Tennessee, a bomb exploded under the house in which Swanner was living in Montgomery, Alabama. The explosion demolished one side of the house, injured Swanner, his wife and two grandchildren, all of whom join as plaintiffs in this action. They allege that the government negligently breached its duty to protect them.

The district court ruled that the government did have a duty to protect Swan-ner and that it breached this duty, but refused to find the government liable because in its opinion plaintiffs did not sustain their burden of showing that the government's negligence was the proximate cause of their injuries. In so ruling the court emphasized the following points:

1. Swanner is a convicted felon.

2. At the time of the bombing, his principal source of income was a part time job repossessing automobiles.

3. Swanner’s landlord was in arrears on the payment of his mortgage and had already been put on notice that foreclosure might be necessary.

4. There was a complete lack of evidence that McGlocklin or his associates either were anywhere near Montgomery at any time or that they had procured the bombing.

While the court properly distinguishes the instant case from Schuster v. City of New York, 5 N.Y.2d 75, 180 N.Y.S.2d 265, 154 N.E.2d 534 (1958) where the victim was a public spirited young man whose murder had no obvious explanation other than retribution for his cooperation with the police, we are impressed by plaintiffs’ contention that they have been held to too strict a standard of proof. We do not endorse the suggestion that the court applied the standard of proof beyond a reasonable doubt. Still, a reading of the court’s opinion in the light of the whole record leaves us with a firm conviction that the plaintiffs were held to a standard of proof stricter than that of a preponderance of the evidence. ^[1]

[*718] To some extent this appears from what the court says explicitly:

“The Court of Appeals of New York suggested in Schuster that it might be possible to create a case of circumstantial evidence so strong as to lead the mind inevitably to the conclusion that injury to a person who supplied information to the police resulted from his having supplied such information. This is not such a case.”

Swanner v. United States, 275 F.Supp. 1007, 1012 (M.D.Ala.1967). We are disturbed by such a formula, particularly the word “inevitably”, especially since no such formula seems to be suggested in Schuster. ^[2]

Moreover, there are other less tangible intimations of the Court’s attitude. For example, inordinate emphasis seems to be attached to the fact that “not one shred of evidence” indicates that Mc-Glocklin or his associates were in or near Montgomery. This assertion, however, impliedly misstates the issue. The real question is whether it is more probable than not that plaintiffs were injured as a result of the disclosures concerning Mc-Glocklin. The very fact that the threat had occurred in the not too distant past and that Swanner was due to testify against McGlocklin just a few days after the bombing occurred we think is powerfully persuasive that there was a causal connection irrespective of any proof that the supposed perpetrators were seen in the area. ^{[2] [3]} Alert avoidance of the classical fallacy of post hoc, ergo propter hoc does not require rejection of common sense inferences.

Had the district court been as sensitive to these considerations as we think it should have been it would perhaps have been more critical in assessing the other “several sources” that could have resulted in the harm to Swanner. While these other sources are certainly possible explanations of the bombing, we think that the court would have scrutinized their likelihood more closely had it not been so occupied with the absence of proof concerning the whereabouts of Mc-Glocklin and his associates on the day in question and had it been more attentive to the proper standard of proof. We see no need to decide whether the court could have reached the same result by applying proper principles because we think that under the circumstances a new trial is required and that the plaintiffs may then be more successful in negating the alternative theories as to the reason for the bombing than they were in this trial.

There is one final issue to be discussed for the guidance of the court at the new trial. Plaintiffs contend that the district court erred in failing to require the production of certain investigative files, principally for the purpose of supplying plaintiffs with the identity of two people who were under suspicion in connection with the bombing. First, it may be stated that the court could properly have granted discovery. Although the government argues that the plaintiffs did not show good cause, it seems that the pertinence of these reports is obvious and in any event cannot be demonstrated by one who does not have access to them. Invoking executive privilege, the government further contends that only the Commissioner of Internal Revenue is authorized to divulge records, see 26 C.F.R.[*719] § 301.9000-1, but this position has been rejected by United States v. Reynolds, 1953, 345 U.S. 1, 73 S.Ct. 528, 97 L.Ed. 727, and more specifically by NLRB v. Capitol Fish Co., 294 F.2d 868, 875, 5 Cir., 1961; accord McFadden v. Avco Corp., 1967, M.D.Ala., 278 F.Supp. 57; Rosee v. Board of Trade, 1964, N.D.Ill, 35 F.R.D. 512. Moreover, while pend-ency of a criminal investigation is a reason for denying discovery of investigative reports, this privilege would not apply indefinitely and it is doubtful that it would apply here so long after the events in question. See Capitol Vending Co. v. Baker, 1964, D.D.C., 35 F.R.D. 510. All in all, it seems clear enough that the government’s claim of privilege does not put the matter beyond the pale of judicial inquiry, especially in matters not affecting national security. See detailed discussion in Timken Roller Bearing Co. v. United States, 1964, N.D.Ohio, 38 F.R.D. 57.

While it would be easy enough to approve discovery had it been granted (any mischief presumably could not be remedied in any event), it is not so clear that we can disapprove the denial of discovery. There is, of course, a broad discretion in the trial court and it is unusual to find abuse of discretion in these matters. See 2A Barron and Holtzoff Federal Practice and Procedure § 657. Here we are dealing with investigative records which if not technically privileged, are at any rate highly sensitive. There would be at least some question as to whether their use in subsequent prosecution is a live issue. Further, it is likely that there would be an admixture of fact and opinion. See O’Keefe v. Boeing Co., 1965, S.D.N.Y., 38 F.R.D. 329, 334. . There would be the possibility, too, that certain matters, e. g., statements by other informers, would be interspersed. All of this would require in camera inspection and disentanglement. In our opinion it was not an abuse of discretion for the district court to decline to get involved in all of this. See Timken, supra, in which it is suggested that this is an area peculiarly liberated from ordinary notions of stare decisis, and where ad hoc policy decisions are justified. Be that as it may, we think the interests of justice would best be served in this case by reversing the judgment of the district court and remanding the case for a new trial.