The plaintiff appeals from the dismissal of her medical malpractice suit under the Federal Tort Claims Act; the district court had ruled that the suit was barred by the Act's two-year statute of limitations for filing the administrative claim that is a prerequisite to suing. The evidence upon which the ruling was based is the plaintiff's deposition, which discloses the following facts. (We do not vouch for their truth; but they are all the record contains.) The wife of an Air Force pilot, Mary Goodhand gave birth to her first and only child on November 29, 1985, at an Air Force hospital. Although the doctors knew that the baby would be unusually large, no arrangements were made to perform a caesarean delivery in the event that a vaginal delivery would prove infeasible or dangerous. After protracted labor, the baby was delivered with the aid of forceps and suction cup, causing Mrs. Goodhand to suffer a fourth-degree perineal tear. In lay terms, as Dr. Reinert, her principal obstetrician--who sewed up the tear--told her, "I had blown out my rectal canal with the delivery, and I would be sore for quite some time."

Shortly after the delivery, Mrs. Goodhand began to experience fecal incontinence. Reinert told her that this was a standard complication of a fourth-degree perineal tear, that the healing process would be slow, but that eventually the tear would heal and the problem abate. He prescribed exercises. She did the exercises but the fecal incontinence continued. She complained to Dr. Reinert continually for the nine months in which both she and the doctor remained at the same base, and he kept assuring her that the problem would clear up eventually. It did not, and she consulted a series of other doctors, who said much the same thing as Reinert. One of them, however, told her in July 1988--more than two and a half years after the delivery of her child--that her incontinence might continue for the rest of her life. Not until February 1990, after a further round of consultations and some unsuccessful surgery, was it finally determined that nothing short of a colostomy would solve the problem, because her internal sphincter was nerve dead. The doctor who told her this was the first who attributed her condition to the repair of the perineal tear rather than to the tear itself. She decided to have a colostomy.

She submitted an administrative tort claim to the Air Force on June 3, 1991. This suit is untimely if her cause of action accrued more than two years before then. 28 U.S.C. Sec. 2401(b). The briefs and record are somewhat unclear as to what the precise act of negligence alleged is, but at the oral argument Mrs. Goodhand's lawyer said that it was the failure of the Air Force doctors, more than five years before she filed her administrative claim, to deliver her baby by caesarean section. Had they done so, she would not have experienced even temporary fecal incontinence, because there would have been no damage to her rectum, let alone the ghastly damage that after years of discomfort and humiliation condemned this young woman to a lifetime of discomfort and humiliation in a different form, as the wearer of a colostomy bag.

The negligent act is not what starts the statute of limitations running, however. It is the discovery of the injury resulting from the negligent act and of the cause of that injury. United States v. Kubrick, 444 U.S. 111, 100 S. Ct. 352, 62 L. Ed. 2d 259 (1979); Winskunas v. Birnbaum, 23 F.3d 1264, 1266-67 (7th Cir.1994); Nemmers v. United States, 795 F.2d 628, 629 (7th Cir.1986). Until the plaintiff knows that he has been injured and by whom or by what, he has no reason to take steps to determine whether he might have a legal claim. This is obvious if he doesn't even know that he has been injured, a distinct possibility in medical malpractice cases because medical interventions can have adverse consequences that do not manifest themselves for months or years or even decades. Even if he knows that he has been injured, if he does not know who or what inflicted the injury, he again has no reason to suppose that he might have a legal claim against someone. This is a common problem in medical malpractice cases, for it is often unclear whether symptoms that appear in the wake of a medical intervention are the consequence of whatever underlying health problem led to the intervention, in which event there is no basis for a legal claim, or of the intervention itself. The same problem arises in legal malpractice cases; a party to a lawsuit may not know whether he lost his case because it was a weak case or because his lawyer was inadequate. Winskunas v. Birnbaum, supra, 23 F.3d at 1266.

Once armed with knowledge that he has been injured and by whom, the potential malpractice plaintiff has reason to believe that he may have a legal claim; and he then has the statutory period in which to conduct the necessary investigation and prepare and file a suit. Mrs. Goodhand acquired the requisite knowledge when, immediately after the delivery, Dr. Reinert told her that she had suffered a torn rectum. That was the injury caused by the Air Force doctors' negligence in failing to deliver the baby by caesarean section. She knew she was injured, and by whom. She had two years in which to determine whether the injury had been a consequence of negligence and to file suit accordingly.

Our analysis has to be qualified in two respects. The first concerns the severity of the injury. The statute of limitations begins to run upon the discovery of the injury, even if the full extent of the injury is not discovered until much later. Fries v. Chicago & Northwestern Transportation Co., 909 F.2d 1092, 1096 (7th Cir.1990); Industrial Constructors Corp. v. U.S. Bureau of Reclamation, 15 F.3d 963, 969 (10th Cir.1994); Manko v. United States, 830 F.2d 831, 842 (8th Cir.1987). This is a general principle of limitations law, not an idiosyncratic feature of the statute of limitations in the Federal Tort Claims Act. Lancaster v. Norfolk & Western Ry., 773 F.2d 807, 821 (7th Cir.1985), and cases cited there.

Were it not for this rule, the statute of limitations might be extended indefinitely--perhaps even to death, since until then it is always possible that the plaintiff's injury will worsen. The statement in Otto v. National Institute of Health, 815 F.2d 985, 989 (4th Cir.1987), that the statute of limitations does not begin to run until the plaintiff "became aware of the true nature of her permanent and irreparable injury" is inconsistent with this rule, see Manko v. United States, supra, 830 F.2d at 842; Burgess v. United States, 744 F.2d 771, 775 n. 9 (11th Cir.1984), and cases cited there, but it is only a dictum. Otto was a case in which the plaintiff was led by the treating physicians to believe that her symptoms were normal complications of measures for treating her disease; it was thus a case of equitable estoppel, of which more shortly.

The rule that the plaintiff cannot wait until the full gravity of his injury is known has an important exception for the case in which at first the injury reasonably seems trivial, and only much later is it discovered to be serious enough to warrant the expense of a precomplaint investigation. Lancaster v. Norfolk & Western Ry., supra, 773 F.2d at 821; Nivens v. Signal Oil & Gas Co., 520 F.2d 1019, 1024, modified on other grounds, 523 F.2d 1382 (5th Cir.1975) (dictum); cf. Hagerty v. L & L Marine Services, Inc., 788 F.2d 315, 317, modified on other grounds, 797 F.2d 256 (5th Cir.1986); Cathcart v. Keene Industrial Insulation, 324 Pa.Super. 123, 471 A.2d 493, 500 n. 10 (1984). We do not want people to file protective suits in cases of trivial injury on the off-chance that the injury may eventually turn out to be serious.

We can find only one example of this principle in Tort Claims land, however, and that a doubtful one. In Burgess v. United States, supra, the delivering physician broke a baby's clavicles to facilitate the delivery, and the baby later developed a paralysis of the upper arms as a consequence. "Trivial" is not quite the word for the injury in Burgess; but maybe it should be understood to mean too slight to induce a reasonable person to undertake the expense of preparing to sue, though we shall not have to decide that in this case. The opinion in Burgess does not mention and may not have been aware of the trivial-injury exception. Instead it treats the palsy as a separate injury that was not known to the baby's parents when the clavicles were broken--that in fact had not yet occurred. But to treat the complications of an injury as a new injury is pretty much to erase the rule that the plaintiff cannot wait to sue until he realizes the full extent of his injury.

Some states appear to have rejected the trivial-injury exception--we ruled in Neubauer v. Owens-Corning Fiberglas Corp., 686 F.2d 570, 575 (7th Cir.1982), that Wisconsin was one--but limitations doctrines in suits under the Federal Tort Claims Act are matters of federal common law, not state law; this is implicit in Kubrick. See also Jastremski v. United States, 737 F.2d 666, 669 (7th Cir.1984); Kossick v. United States, 330 F.2d 933 (2d Cir.1964). Despite the absence of examples besides the dubious Burgess, and the rejection of the exception by the Fifth Circuit in Jones Act cases, notably Crisman v. Odeco, Inc., 932 F.2d 413, 415 (5th Cir.1991) (possibly misreading the Hagerty opinion, supra ), we cannot think of any reason not to apply the trivial-injury exception to cases under the Tort Claims Act.

The second qualification to the rule that the statute of limitations in the Act begins to run when the victim of the tort knows that he has been injured and by whom concerns lulling efforts by defendants, as in the Otto case that we discussed earlier. If the defendants make efforts to discourage the victim from suing or from instigating investigations that might lead to suit, they may be estopped to plead the statute of limitations if the victim acts reasonably in light of the situation as it has been made to appear to him by the defendants. Singletary v. Continental Illinois Nat'l Bank, 9 F.3d 1236, 1241 (7th Cir.1993). Although the extent to which doctrines of estoppel apply to the federal government remains unsettled, Citizens Marine Nat'l Bank v. Department of Commerce, 854 F.2d 223, 229 (7th Cir.1988); Crawford v. United States, 796 F.2d 924, 926 (7th Cir.1986); Zavala by Ruiz v. United States, 876 F.2d 780, 783 (9th Cir.1989), the Supreme Court has now held that there is a "rebuttable presumption" that equitable tolling is a defense to all federal statutes of limitations. Irwin v. Dept. of Veterans Affairs, 498 U.S. 89, 95-96, 111 S. Ct. 453, 457-458, 112 L. Ed. 2d 435 (1990). Irwin was not a Tort Claims case, and equitable tolling is not the same thing as equitable estoppel. Central States, Southeast & Southwest Areas Pension Fund v. Navco, 3 F.3d 167, 174 (7th Cir.1993); Cada v. Baxter Healthcare Corp., 920 F.2d 446, 450-51 (7th Cir.1990). But Irwin explicitly (albeit in dictum) treats equitable estoppel as a form of equitable tolling. 498 U.S. at 96, 111 S.Ct. at 457. All the cases to address the question whether equitable estoppel is a defense to the Tort Claims Act's statute of limitations hold that it is (though there are none in this circuit). See, e.g., Industrial Constructors Corp. v. U.S. Bureau of Reclamation, supra, 15 F.3d at 969; Zavala by Ruiz v. United States, supra, 876 F.2d at 783; Chamness v. United States, 835 F.2d 1350, 1353 (11th Cir.1988); cf. Krueger v. Saiki, 19 F.3d 1285 (8th Cir.1994) (per curiam). We cannot think of any reason why if equitable tolling is a defense, equitable estoppel should not be. Nor can we think of any basis in the Federal Tort Claims Act for rebuttal of the presumption that equitable tolling (including equitable estoppel) should be a defense. But neither equitable estoppel nor the trivial-injury rule can help Mrs. Goodhand so far as her claim based on the consequences of the tear is concerned.

Long before June 3, 1989, the last possible date on which her cause of action could have accrued without her suit being untimely, she knew that she had suffered a serious injury. For by then she had been suffering from fecal incontinence for two and a half years. We know that chronic fecal incontinence is a serious condition, because it drove her to obtain a colostomy. The fact that not until 1990 were the full dimensions of the problem known--until then there was hope that the problem could be solved without a colostomy--is irrelevant to the running of the statute of limitations. Only if the injury is initially believed (reasonably) to be too slight to warrant the expense, inconvenience, and uncertainties of litigation is the running of the statute of limitations suspended until the injury is discovered to be serious. It was obvious that Mrs. Goodhand's injury was nontrivial when her fecal incontinence did not abate after a few weeks.

So even if the Air Force doctors could be accused of trying to minimize the severity of her condition, there would be no basis for an estoppel, because a plaintiff cannot wait to sue until she learns the full extent of her injury. It is not as if the doctors had pretended that she did not have fecal incontinence--that she had not been injured--or that the injury was trivial, not worth suing over. Although they did not tell her that it was a permanent condition, there is no suggestion that they knew, and even the impermanent form was not trivial. The doctors did not conceal the cause of the injury (at least if the cause was the form of delivery rather than the way in which the tear was repaired--the significance of this qualification will become clear in a moment). Dr. Reinert told Mrs. Goodhand that the problem had arisen from the form of delivery, vaginal rather than caesarean. He did not tell her that he had been negligent in not having a doctor standing by in the delivery room who could deliver the baby by caesarean section in the not unlikely eventuality (because of the size of the baby) that vaginal delivery threatened to injure Mrs. Goodhand. But the statute of limitations does not begin to run on the date when the plaintiff discovers or should have discovered that the defendant has been negligent. That was the position rejected in Kubrick. It begins to run on the date when the plaintiff discovers that he has been injured by an act or omission attributable to the defendant. The plaintiff then has the statutory period to determine whether the act or omission was negligent, and to proceed from there.

There is a further wrinkle in this case, however. We said that the alleged act of negligence was the form of delivery, that is, vaginal rather than caesarean. That certainly was one alleged act of negligence, but not the only one; and the fact that in the heat of argument Mrs. Goodhand's lawyer did not mention any other should not be held against him. The complaint, not yet superseded because there has been no discovery directed to the merits of the suit (the defense of statute of limitations, being jurisdictional in a suit under the Tort Claims Act, must be adjudicated definitively before the case proceeds to the merits, Crawford v. United States, supra, 796 F.2d at 927-29), alleges another act of negligence: negligent repair of the perineal tear. At some point in her round of physician consultations, a reasonable person in Mrs. Goodhand's position might have begun to suspect that fecal incontinence of so protracted and stubborn a character was not just the "normal" result of a fourth-degree perineal tear, but must have something to do with its repair by Dr. Reinert. No one has attempted to determine when that point was reached; and one might be inclined to ask whether, if the doctors didn't suspect that the repair rather than the tear was the cause of the injury, she should have. She was finally told in February 1990, but that was well within the statutory period for suit; maybe, however, she should have suspected the repair much earlier, which might make the suit untimely. Or might not. A relevant consideration is that Dr. Reinert's attribution of Mrs. Goodhand's problem to the tear, without mention that it might have really been the repair that was responsible, could be--we do not say it should be; that is a matter for consideration by the district court in the first instance--regarded as the kind of lulling statement that extends the period for bringing suit; but even so, again we don't yet know for how long.

So we may have the unusual case of two alleged acts of negligence, occurring in rapid sequence (the vaginal delivery, and the repair of the tear caused by the delivery) committed by the same person on the same person, yet only one of the acts starts the statute of limitations running. The question is whether, when suit based on that act is time barred, suit based on the other is barred too. We think not. Although Dr. Reinert both failed to arrange for a caesarean delivery and repaired (perhaps negligently) the consequences of his mistake, we cannot think of a reason why the case should be decided differently than if two different doctors had been responsible for the two mistakes, the two injuries. Then certainly the running of the statute of limitations against one for his act would not bar a suit against the other based on the other's act. Raddatz v. United States, 750 F.2d 791, 795-96 (9th Cir.1984). It would be extremely odd if by committing two negligent acts rather than one, a defendant obtained an immunity that would be denied two defendants each of whom committed just one of the two negligent acts and thus was, as it were, only half as negligent. We do not think the single defendant would be immune.

Since it is unclear from the record when Mrs. Goodhand in the exercise of due diligence should have discovered that she had been injured (if she had been--we have only allegations at this stage) by Dr. Reinert's repair of the perineal tear, as distinct from being injured by the tear itself (and hence by the choice of vaginal delivery), which he told her was the injury, the case must be remanded for a determination by the district judge whether suit based on the repair is time barred. If not, the case can proceed, although limited to negligence in the repair. To prevail under that theory of negligence, Mrs. Goodhand will have to prove not only that the repair was negligent but also that it was the cause of the problem of fecal incontinence that led her to undergo the colostomy, or at least that it aggravated her condition to an extent that can reasonably be quantified in an award of damages.

To assure consistency with Kubrick, we remind that the question is not when Mrs. Goodhand learned or should have learned that her fecal incontinence was the result of negligence in the repair of the perineal tear. It was when she learned that it was the result (in whole or part) of the repair, whether negligent or not, as distinct from the tear itself. The statute of limitations begins to run when the plaintiff discovers the cause of the injury, not when he discovers either that he was injured, which usually comes earlier, or that the injurer was negligent, which usually comes later. We discussed these distinctions in Drazan v. United States, 762 F.2d 56 (7th Cir.1985), which we commend to the district judge and the parties for guidance on remand.

The judgment of dismissal is vacated and the case remanded for further proceedings consistent with this opinion.

VACATED AND REMANDED.