No. 97-6057, 97-6099.United States Court of Appeals, Second Circuit.Argued: October 22, 1997.
Decided: March 20, 1998.
Appeal and cross-appeal from a judgment of the United States District Court for the District of Connecticut (Warren W. Eginton, Judge) holding the United States liable for negligence under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680, and awarding the plaintiff $1,034,236.02 in damages for the death of his wife.
MARY JO DONAHUE, Trial Attorney, Torts Branch, Civil Division, U.S. Department of Justice, Washington, D.C. (Frank W. Hunger, Assistant Attorney General; Christopher F. Droney, United States Attorney; Carl J. Schuman, Assistant United States Attorney; Jeffrey Axelrad, Director, Torts Branch; Roger D. Einerson, Assistant Director, Torts Branch, on the brief), for Defendant-Appellant Cross-Appellee.
ROBERT I. REARDON, JR., The Reardon Law Firm, P.C., New London, CT (Angelo A. Ziotas, on the brief), for Plaintiff-Appellee Cross-Appellant.
Before: NEWMAN, ALTIMARI, and CALABRESI, Circuit Judges.
CALABRESI, Circuit Judge:
 The defendant, the United States of America, appeals from a judgment of the United States District Court for the District of Connecticut (Warren W. Eginton, Judge). This suit under the Federal Tort Claims Act, 28 U.S.C. §§ 1346(b), 2671-2680, was originally filed by Patricia Zuchowicz, who claimed to have developed primary pulmonary hypertension, a fatal lung condition, as a result of the defendant’s negligence in prescribing an overdose of the drug Danocrine. Following Mrs. Zuchowicz’s death in 1991, her husband, Steven, continued the case on behalf of his wife’s estate, claiming that the defendant was responsible for her death. After a bench trial, the district court awarded the plaintiff $1,034,236.02 in damages. [JA-053]
 The case statement recited above goes to the heart of the law of torts. A plaintiff claims to have developed a fatal condition as a result of a defendant’s negligence in prescribing an excessive amount of a drug — a tragic injury allegedly caused by defendant’s wrong. There is no doubt in the case before us either as to the injury or as to the defendant’s wrong; both are conceded. The only issue is causation.
 Did the action for which the defendant is responsible cause, in a legal sense, the harm which the plaintiff suffered? — a question easily put and often very hard to answer. There is, moreover, no older requirement in this area of law than the need to show such a link between the defendant’s actions and the plaintiff’s loss. It long precedes the obligation to show that the defendant was at fault. Along with the showing of injury,
causation constituted an essential part of what the plaintiff had to demonstrate for the early common law action in trespass to lie.
 Over the centuries the courts have struggled to give meaning to this requirement — in the simplest of situations, who hit whom, and in the most complex ones, which polluter’s emissions, if any, hurt which plaintiff. It is the question that we must seek to answer today in the context of modern medicine and a very rare disease.
 I. Background A. Drug, Illness, and Death 1. The Overdose
 The facts, as determined by the district court, are as follows. On February 18, 1989, Mrs. Zuchowicz filled a prescription for the drug Danocrine at the Naval Hospital pharmacy in Groton, Connecticut. The prescription erroneously instructed her to take 1600 milligrams of Danocrine per day, or twice the maximum recommended dosage. The defendant has stipulated that its doctors and/or pharmacists were negligent and violated the prevailing standard of medical care by prescribing this wrong dosage.
 Mrs. Zuchowicz took the 1600 milligrams of Danocrine each day for the next month. Thereafter, from March 24 until May 30, she took 800 milligrams per day. While taking Danocrine she experienced abnormal weight gain, bloating, edema, hot flashes, night sweats, a racing heart, chest pains, dizziness, headaches, acne, and fatigue. [JA033-34] On May 30, she was examined by an obstetrician/gynecologist in private practice who told her to stop taking the Danocrine. During the summer, she continued to experience severe fatigue and chest tightness and pain, and began having shortness of breath. In October 1989, she was diagnosed with primary pulmonary hypertension (“PPH”), a rare and fatal disease in which increased pressure in an individual’s pulmonary artery causes severe strain on the right side of the heart. [JA034] At the time she was diagnosed with the disease, the median life expectancy for PPH sufferers was 2.5 years. Treatments included calcium channel blockers and heart and lung transplantation.
 Mrs. Zuchowicz was on the waiting list for a lung transplant when she became pregnant. Pregnant women are not eligible for transplants, and pregnancy exacerbates PPH. Mrs. Zuchowicz gave birth to a son on November 21, 1991. She died one month later, on December 31, 1991. [JA036]
 2. Primary Pulmonary Hypertension
 Pulmonary hypertension is categorized as “primary” when it occurs in the absence of other heart or lung diseases. “Secondary” pulmonary hypertension is diagnosed when the hypertension results from another heart or lung disease, such as emphysema or blood clots. PPH is very rare. A National Institute of Health registry recorded only 197 cases of PPH from the mid-1980s until 1992. [Red Br. at 5-6] It occurs predominantly in young women. Exogenous agents known to be capable of causing PPH include birth control pills, some appetite suppressants, chemotherapy drugs, rapeseed oil, and L-Tryptophan. [JA034]
 According to the district court’s findings of fact, the disease involves the interplay of the inner layers of the pulmonary blood vessels known as the endothelium and the vascular smooth muscle. The endothelium releases substances called vasodilators and vasoconstrictors, which dilate and constrict the blood vessels. These substances can also cause growth of the vascular smooth muscle. Experts currently believe that an imbalance in
vasodilators and vasoconstrictors plays a part in the development of pulmonary hypertension. If too many vasoconstrictors are released, the blood vessels contract, the endothelial cells die, and the vascular smooth muscle cells proliferate. These actions create increased pulmonary vascular resistance. [JA035]
 3. Danocrine
 Danocrine has been extensively studied and prescribed since the late 1960s for endometriosis. According to the testimony of plaintiff’s expert Dr. W. Paul D’Mowski, who personally performed much of the initial research on the drug, Danocrine is safe and effective when administered properly. Based on studies by Dr. D’Mowski and others, Danocrine was approved by the Food and Drug Administration (“FDA”) for use in dosages not to exceed 800 mg/day. Mrs. Zuchowicz was accidentally given a prescription instructing her to take twice this amount — 1600 mg/day. According to Dr. D’Mowski no formal studies of the effects of Danocrine at such high doses have been performed, and very, very few women have received doses this high in any setting.
 B. The Expert Testimony
 The rarity of PPH, combined with the fact that so few human beings have ever received such a high dose of Danocrine, obviously impacted on the manner in which the plaintiff could prove causation. The number of persons who received this type of overdose was simply too small for the plaintiff to be able to provide epidemiological, or even anecdotal, evidence linking PPH to Danocrine overdoses. The plaintiff (Mrs. Zuchowicz’s husband and executor), therefore, based his case primarily on the testimony of two expert witnesses, Dr. Richard Matthay, a physician and expert in pulmonary diseases, and Dr. Randall Tackett, a professor of pharmacology who has published widely in the field of the effects of drugs on vascular tissues. In rendering a judgment for the plaintiff, the district court relied heavily on the evidence submitted by these two experts. The defendant challenges both the admissibility and the sufficiency of their testimony.
 1. Dr. Matthay
 Dr. Richard Matthay is a full professor of medicine at Yale and Associate Director and Training Director of Yale’s Pulmonary and Critical Care Section. [JA035] He is a nationally recognized expert in the field of pulmonary medicine, with extensive experience in the area of drug-induced pulmonary diseases. Dr. Matthay examined and treated Mrs. Zuchowicz. His examination included taking a detailed history of the progression of her disease, her medical history, and the timing of her Danocrine overdose and the onset of her symptoms.
 Dr. Matthay testified that he was confident to a reasonable medical certainty that the Danocrine caused Mrs. Zuchowicz’s PPH. When pressed, he added that he believed the overdose of Danocrine to have been responsible for the disease. His conclusion was based on the temporal relationship between the overdose and the start of the disease and the differential etiology method of excluding other possible causes. While Dr. Matthay did not rule out all other possible causes of pulmonary hypertension, he did exclude all the causes of secondary pulmonary hypertension. On the basis of Mrs. Zuchowicz’s history, he also ruled out all previously known drug-related causes of primary pulmonary hypertension.
 Dr. Matthay further testified that the progression and timing of Mrs. Zuchowicz’s disease in relation to her overdose supported a finding of drug-induced PPH. Dr. Matthay emphasized that, prior to the overdose, Mrs. Zuchowicz was a healthy, active young woman with no history of cardiovascular problems, and that, shortly after the overdose, she began experiencing symptoms of PPH such as weight gain, swelling of hands and feet, fatigue, and shortness of breath. He described the similarities between the course of Mrs. Zuchowicz’s illness and that of accepted cases of drug-induced PPH, and he went on to discuss cases involving classes of drugs that are known to cause other pulmonary diseases (mainly anti-cancer drugs). He noted that the onset of these diseases, which are recognized to be caused by the particular drugs, was very similar in timing
and course to the development of Mrs. Zuchowicz’s illness. [JA039-040]
 2. Dr. Tackett
 Dr. Randall Tackett is a tenured, full professor of pharmacology and former department chair from the University of Georgia. [JA036] He has published widely in the field of the effects of drugs on vascular tissues. Dr. Tackett testified that, to a reasonable degree of scientific certainty, he believed that the overdose of Danocrine, more likely than not, caused PPH in the plaintiff by producing: 1) a decrease in estrogen; 2) hyperinsulinemia, in which abnormally high levels of insulin circulate in the body; and 3) increases in free testosterone and progesterone. Dr. Tackett testified that these hormonal factors, taken together, likely caused a dysfunction of the endothelium leading to PPH. Dr. Tackett relied on a variety of published and unpublished studies that indicated that these hormones could cause endothelial dysfunction and an imbalance of vasoconstrictor effects. [JA040]
 II. Discussion A. Was the Admission of the Plaintiff’s Experts’ Testimony Manifestly Erroneous?
 The defendant’s first argument is that the district court erred in admitting the testimony of Dr. Tackett and Dr. Matthay. We review the district court’s decision to admit or exclude expert testimony under a highly deferential abuse of discretion standard. See General Elec. Co. v. Joiner, 118 S.Ct. 512, 517 (1997); McCullock v. H.B. Fuller Co., 61 F.3d 1038, 1042 (2d Cir. 1995) (“The decision to admit expert testimony is left to the broad discretion of the trial judge and will be overturned only when manifestly erroneous.”).
 The Federal Rules of Evidence permit opinion testimony by experts when the witness is “qualified as an expert by knowledge, skill, experience, training, or education,” and “[i]f scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue.” Fed. R. Evid. 702. And though in Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 588-89 (1993), the Supreme Court altered the traditional test for the admissibility of expert testimony, it did not change the standard of appellate review of these decisions, see Joiner, 118 S.Ct. at 517.
 Under Daubert, trial judges are charged with ensuring that expert testimony “both rests on a reliable foundation and is relevant to the task at hand.” 509 U.S. at 597. Thus, while Daubert and the Federal Rules of Evidence “allow district courts to admit a somewhat broader range of scientific testimony than would have been admissible under Frye, they leave in place the `gatekeeper’ role of the trial judge in screening such evidence.” Joiner, 118 S.Ct. at 517. Indeed Daubert strengthens this role, for it requires that judges make a “preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid and of whether that reasoning or methodology properly can be applied to the facts in issue.” 509 U.S. at 592-93.
 The factors identified by the Supreme Court as relevant to this inquiry are: (1) whether the theory can be (and has been) tested according to the scientific method; (2) whether the theory or technique has been subjected to peer review and publication; (3) in the case of a particular scientific technique, the known or potential rate of error; and (4) whether the theory is generally accepted. See id. at 593-94. The Court emphasized, however, that these factors were not an exclusive or dispositive list of what should be considered, and
that the trial court’s inquiry should be a “flexible one.” Id. at 594.
 The question in this case is whether, in light of these factors, the district court’s decision to admit the testimony of Dr. Matthay and Dr. Tackett was an abuse of discretion. We addressed a similar question in McCullock v. H.B. Fuller Co., 61 F.3d 1038 (2d Cir. 1995). In McCullock, we upheld the district court’s decision to admit the testimony of an engineer and a medical doctor in a case involving a worker’s exposure to glue fumes and her subsequent development of throat polyps. Applying the “manifestly erroneous” standard, we rejected the defendant’s argument that the district court had not properly performed its gatekeeping function as required by Daubert. See id. at 1042-44. With respect to the doctor’s testimony, we noted that the doctor based his opinion on a range of factors, including his care and treatment of [the plaintiff]; her medical history . . .; pathological studies; . . . his training and experience; use of a scientific analysis known as differential etiology (which requires listing possible causes, then eliminating all causes but one); and reference to various scientific and medical treatises.
 Id. at 1044. And we pointed out that the “[d]isputes as to the strength of his credentials, faults in his use of differential etiology as a methodology, or lack of textual authority for his opinion, go to the weight, not the admissibility of his testimony.” Id.
 McCullock provides strong support for the instant plaintiff’s position. In the case before us, as in McCullock, the district court carefully undertook and fulfilled its role in making the evaluation required by Daubert — a “preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid and of whether that reasoning or methodology properly can be applied to the facts in issue.” 509 U.S. at 592-93. Where, as in this case, the district court decides to admit the testimony of well-credentialed experts relying on scientific methodology, we should and will be reluctant to upset that decision as an abuse of discretion.
 In the district court, the defendant made substantially the same arguments, regarding the validity of the methods used by Dr. Matthay and Dr. Tackett in reaching their conclusions, that it now raises on appeal. The district court rejected these arguments, stating that the plaintiff’s experts “based their opinions on methods reasonably relied on by experts in their particular fields.” [JA039] We do not believe that the district court’s decision in this regard was erroneous, let alone manifestly so.
 B. Were the District Court’s Factual Findings with Respect to Causation Clearly Erroneous?
 We review the district court’s factual findings for clear error. See, e.g., Mathie v. Fries, 121 F.3d 808, 811 (2d Cir. 1997); see also Fed. R. Civ. P. 52(a). The defendant argues that, even assuming that the testimony of the plaintiff’s experts was admissible, the district court’s finding that the Danocrine overdose more likely than not caused Mrs. Zuchowicz’s illness was clearly erroneous. The defendant contends that, since Danocrine has never been previously linked to PPH, the district court’s conclusion that the drug caused Mrs. Zuchowicz’s illness was impermissible. For the reasons stated below, we reject the defendant’s arguments.
 1. Applicable Law
 The liability of the federal government under the Federal Tort Claims Act is determined according to the law of the state in which the injury occurred. See 28 U.S.C. § 1346(b); Taylor v. United States, 121 F.3d 86, 89 (2d Cir. 1997). Connecticut law, therefore, provides the applicable standards in this case. A plaintiff alleging medical malpractice in Connecticut must first prove that the defendant negligently deviated from the customary standard of care. See, e.g., Edwards v. Tardif, 692 A.2d 1266, 1269 (Conn. 1997). Since the defendant has stipulated that its agents were negligent in prescribing an overdose of the drug Danocrine, there is no question that this requirement is satisfied. In addition, “the plaintiff must establish a causal relationship between the physician’s negligent actions or failure to act and the resulting injury by showing that the action or omission constituted a substantial factor in
producing the injury.” Id.; see also Mather v. Griffin Hosp., 540 A.2d 666, 669 (Conn. 1988). This “substantial factor” causation requirement is the crux of the case before us.
 2. The Connecticut Law of Causation
 To meet the requirement that defendant’s behavior was a substantial factor in bringing about the plaintiff’s injury, the defendant must generally show:
 (a) that the defendant’s negligent act or omission was a but for cause of the injury,
 (b) that the negligence was causally linked to the harm,
 (c) that the defendant’s negligent act or omission was proximate to the resulting
 This has long been the law in Connecticut, where Chief Justice Maltbie in Kinderavich v. Palmer, 15 A.2d 83 (Conn. 1940), the leading Connecticut discussion of causation, expressly noted all three requirements decades before they were fully distinguished in the scholarly literature. See id. at 86-87.
 In criticizing the use of an all-encompassing terminology of proximate cause, Maltbie separated out “those situations where the accident . . . would have happened whether or not the act or omission in question had occurred.” Id. at 86 (discussing but for cause). He also distinguished those cases where the “particular conduct claimed to be negligent . . . `had no real connection’ with the injuries suffered.” Id. (discussing causal link).
And, finally, having differentiated these situations — and some others not currently germane — from those involving the requirement of proximate cause, the late Chief Justice went on to discuss at some length how proximity of causation was to be determined in Connecticut. See id. at 87-89.
 In the case before us, as we shall see, neither the requirement of proximity nor that of causal link gives rise to any problems (though the presence of a strong causal link will prove to be highly significant). The case turns only on the difficulty of showing a but for cause. On whether, in other words, the plaintiff has sufficiently demonstrated: (a) that defendant’s act in giving Mrs. Zuchowicz Danocrine was the source of her illness and death, and (b) that it was not just the Danocrine, but its negligent overdose that led to Mrs. Zuchowicz’s demise.
 3. Connecticut Law and Experts
 In seeking to show both components of but for causation, plaintiff’s reliance on experts must meet the substantive requirements of Connecticut law. Under that law, “[t]he expert opinion that seeks to establish the causal connection between the injury and the alleged negligence `must rest upon more than surmise or conjecture.'” Shelnitz v. Greenberg, 509 A.2d 1023, 1027 (Conn. 1986) (citation omitted); see also Aspiazu v. Orgera, 535 A.2d 338, 342 (Conn. 1987). The expert must deal not in mere possibilities, but in “reasonable medical probabilit[ies].” Aspiazu, 535 A.2d at 342. At the same time, it is well-established that causation “may be proved by circumstantial evidence,” Shelnitz, 509 A.2d at 1027 (internal quotations and citation omitted), and that “[t]he causal relation between an injury and its later physical effects may be established by the direct opinion of a physician, by his deduction by the process of eliminating causes other than the traumatic agency, or by his opinion based upon a hypothetical question,” id. at 1028 (internal quotation marks and citation omitted).
 4. Was Danocrine a But For Cause of Mrs. Zuchowicz’s Illness and Death?
 We hold that, on the basis of Dr. Matthay’s testimony alone, the finder of fact
could have concluded — under Connecticut law — that Mrs. Zuchowicz’s PPH was, more likely than not, caused by Danocrine. While it was not possible to eliminate all other possible causes of pulmonary hypertension, the evidence presented showed that the experts had not only excluded all causes of secondary pulmonary hypertension, but had also ruled out all the previously known drug-related causes of PPH. In addition, Dr. Matthay testified, based on his expertise in pulmonary diseases, that the progression and timing of Mrs. Zuchowicz’s illness in relationship to the timing of her overdose supported a finding of drug-induced PPH to a reasonable medical certainty. In this respect, we note that in the case before us, unlike many toxic torts situations, there was not a long latency period between the onset of symptoms and the patient’s exposure to the drug that was alleged to have caused the illness. Rather, as Dr. Matthay testified, the plaintiff began exhibiting symptoms typical of drug-induced PPH shortly after she started taking the Danocrine. Under the circumstances, we cannot say that the fact finder was clearly erroneous in determining that, more probably than not, the Danocrine caused Mrs. Zuchowicz’s illness.
 5. Was the Overdose a But For Cause of Mrs. Zuchowicz’s Illness and Death?
 To say that Danocrine caused Mrs. Zuchowicz’s injuries is only half the story, however. In order for the causation requirement to be met, a trier of fact must be able to determine, by a preponderance of the evidence, that the defendant’s negligence was responsible for the injury. In this case, defendant’s negligence consisted in prescribing an overdose of Danocrine to Mrs. Zuchowicz. For liability to exist, therefore, it is necessary that the fact finder be able to conclude, more probably than not, that the overdose was the cause of Mrs. Zuchowicz’s illness and ultimate death. The mere fact that the exposure to Danocrine was likely responsible for the disease does not suffice.
 The problem of linking defendant’s negligence to the harm that occurred is one that many courts have addressed in the past. A car is speeding and an accident occurs. That the car was involved and was a cause of the crash is readily shown. The accident, moreover, is of the sort that rules prohibiting speeding are designed to prevent. But is this enough to support a finding of fact, in the individual case, that speeding was, in fact, more probably than not, the cause of the accident? The same question can be asked when a car that was driving in violation of a minimum speed requirement on a super-highway is rear-ended. Again, it is clear that the car and its driver were causes of the accident. And the accident is of the sort that minimum speeding rules are designed to prevent. But can a fact finder conclude, without more, that the driver’s negligence in driving too slowly led to the crash? To put it more precisely — the defendant’s negligence was strongly causally linked to the accident, and the defendant was undoubtedly a but for cause of the harm, but does this suffice to allow a fact finder to say that the defendant’s negligence was a but for cause?
 At one time, courts were reluctant to say in such circumstances that the wrong could be deemed to be the cause. They emphasized the logical fallacy of post hoc, ergo propter hoc, and demanded some direct evidence connecting the defendant’s wrongdoing to the harm. See, e.g., Wolf v. Kaufmann, 227 A.D. 281, 282, 237 N.Y.S. 550, 551 (App. Div. 1929) (denying recovery for death of plaintiff’s decedent, who was found unconscious at foot of stairway which, in violation of a statute, was unlighted, because the plaintiff had offered no proof of “any causal connection between the accident and the absence of light”).
 All that has changed, however. And, as is so frequently the case in tort law, Chief Judge Cardozo in New York and Chief Justice Traynor in California led the way. In various opinions, they stated that: if (a) a negligent act was deemed wrongful because that act increased the chances that a particular type of accident would occur, and (b) a mishap of that very sort did happen, this was enough to support a finding by the trier of fact that the negligent behavior caused the harm. Where such a strong causal link exists, it is up to the negligent party to bring in evidence denying but for cause and suggesting
that in the actual case the wrongful conduct had not been a substantial factor.
 Thus, in a case involving a nighttime collision between vehicles, one of which did not have the required lights, Judge Cardozo stated that lights were mandated precisely to reduce the risk of such accidents occurring and that this fact sufficed to show causation unless the negligent party demonstrated, for example, that in the particular instance the presence of very bright street lights or of a full moon rendered the lack of lights on the vehicle an unlikely cause. See Martin v. Herzog, 126 N.E. 814, 816 (N.Y. 1920); see also Clark v. Gibbons, 426 P.2d 525, 542 (Cal. 1967) (Traynor, C.J., concurring in part and dissenting in part on other grounds).
 The general acceptance of this view is both signaled and explained by Prosser, which states categorically:
 And whether the defendant’s negligence consists of the violation of some statutory safety regulation, or the breach of a plain common law duty of care, the court can scarcely overlook the fact that the injury which has in fact occurred is precisely the sort of thing that proper care on the part of the defendant would be intended to prevent, and accordingly allow a certain liberality to the jury in drawing its conclusion.
 Prosser, supra note 6, § 41, at 270; see also Calabresi, supra note 6, at 71-73.
 It is clear that Connecticut accepts this approach. See, e.g., Knybel v. Cramer, 29 A.2d 576, 577-78 (Conn. 1942) (after asking whether the defendant’s negligence was the cause of an injury, the Connecticut Supreme Court of Errors stated “[w]here a statute is designed to protect persons against injury, one who has, as a result of its violation, suffered such an injury as the statute was intended to guard against has a good ground of recovery.”); see also Small v. South Norwalk Savs. Bank, 535 A.2d 1292, 1296 (Conn. 1988).
 The case before us is a good example of the above-mentioned principles in their classic form. The reason the FDA does not approve the prescription of new drugs at above the dosages as to which extensive tests have been performed is because all drugs involve risks of untoward side effects in those who take them. Moreover, it is often true that the higher the dosage the greater is the likelihood of such negative effects. At the approved dosages, the benefits of the particular drug have presumably been deemed worth the risks it entails. At greater than approved dosages, not only do the risks of tragic side effects (known and unknown) increase, but there is no basis on the testing that has been performed for supposing that the drug’s benefits outweigh these increased risks. See generally 21 U.S.C. § 355(d) (indicating that the FDA should refuse to approve a new drug unless the clinical tests show that the drug is safe and effective for use under the conditions “prescribed, recommended, or suggested in the proposed labeling”). It follows that when a negative side effect is demonstrated to be the result of a drug, and the drug was wrongly prescribed in an unapproved and excessive dosage (i.e. a strong causal link has been shown), the plaintiff who is injured has generally shown enough to permit the finder of fact to conclude that the excessive dosage was a substantial factor in producing the harm.
 In fact, plaintiff’s showing in the case before us, while relying on the above stated principles, is stronger. For plaintiff introduced some direct evidence of causation as well. On the basis of his long experience with drug-induced pulmonary diseases, one of plaintiff’s experts, Dr. Matthay, testified that the timing of Mrs. Zuchowicz’s illness led him to conclude that the overdose (and not merely Danocrine) was responsible for her catastrophic reaction.
 Under the circumstances, we hold that defendant’s attack on the district court’s finding of causation is meritless.
 C. Damages 1. Defendant’s Argument
 The defendant claims that the district court’s award of damages for lost wages and earning capacity was not supported by the record. Prior to trial, the defendant suggested, and the plaintiff agreed
to, a proposed finding of fact as to the amount of Mrs. Zuchowicz’s earnings from her work as nurse’s aide in 1987 and 1988 ($4301 and $5284, respectively). The defendant now objects to the district court’s use of these numbers in calculating Mrs. Zuchowicz’s lost earnings. The defendant seems to have overlooked the elementary principle of trial practice that once a fact has been agreed to by both parties and, as a result of such agreement has been submitted to the trial court as a proposed finding of fact, it need not be proved at trial. (And this remains so regardless of whether the parties have formally termed such a proposed finding a “stipulation”). The purpose of a pre-trial stipulation of this sort is precisely to narrow the scope of trial by eliminating issues that the parties do not dispute. It follows that the defendant’s argument with respect to damages is completely without merit.
 2. Plaintiff’s Request for Additur
 On cross-appeal, the plaintiff contends that the $900,000 in non-economic damages awarded by the district court was insufficient, claiming that this amount of money is so small in comparison to the harm suffered by Mrs. Zuchowicz that it must be overturned. We find that the district court’s damage award was well within the range of appropriate awards, and reject the plaintiff’s request.
 III. Conclusion
 We have examined all of the defendant’s arguments and find them to be without merit. Accordingly, the judgment of the district court is affirmed.
In the last fifty years the strictness of the requirement that the plaintiff show that without defendant’s act or omission the accident would not have occurred has been mitigated in several types of cases. For instance, where two defendants are both clearly at fault, where the plaintiff has little or no information as to which one’s negligence was responsible for the injury, and especially where the defendants may have better access to such information, the modern trend is to place the burden on the defendants to disprove causation. See, e.g., Summers v. Tice, 199 P.2d 1, 4 (Cal. 1948); see also, Modave v. Long Island Jewish Med. Ctr., 501 F.2d 1065, 1072-74 (2d Cir. 1974) (Friendly, J.) (suggesting that, under New York law, a plaintiff may not need to prove which of two culpable defendants actually caused the plaintiff’s injury even when the defendants were probably no more able to show what happened than was the plaintiff). Another important example of this easing trend has been the acceptance of statistical or market share evidence as a means of assigning at least part of a loss to various defendants whose conduct justified liability but who could not be identified, more probably than not, as having been but for causes of it. See, e.g., Sindell v. Abbott Labs., 607 P.2d 924, 936-37 (Cal. 1980); Hymowitz v. Eli Lilly Co., 539 N.E.2d 1069, 1075 (N.Y. 1989). Many courts long ago abandoned the requirement of but for cause in situations where, since the negligence of any one of several defendants was sufficient to cause the harm, the negligence of none was its necessary cause. See, e.g., Corey v. Havener, 65 N.E. 69, 69 (Mass. 1902). Indeed, some commentators attribute the acceptance of the “substantial factor” terminology, such as that used in Connecticut, to the problems a strict but for test would cause in this latter type of case. See, e.g., W. Page Keeton et al., Prosser and Keeton on the Law of Torts § 41, at 267-68 (5th ed. 1984) [hereinafter Prosser]. While none of these easings in the requirement of proof of but for cause applies directly to the case before us, it is not unlikely that developments that are relevant to the instant case, see infra section II(B)(4), derived from a desire to achieve analogous goals through tort law. See generally Guido Calabresi, Concerning Cause and the Law of Torts: An Essay for Harry Kalven, Jr., 43 U. Chi. L. Rev. 69 (1975).
The leading case involving this requirement is Berry v. Sugar Notch Borough, 43 A. 240 (Pa. 1899). In Berry, a tree fell on a trolley car whose excess speed had caused the tram to be at that specific place when the tree fell. The court held that the requirement of causation was not met. This result was correct since, although the accident would not have occurred but for the trolley’s speeding, speeding does not increase the probability of trees falling on trolleys. Other similar cases (termed “darting out” cases) involve speeders who but for their velocity would not have been at the particular spot when children darted out from behind trees, etc., and were hit. In such cases — assuming that, had the speeders been at the same spot at the same time, they would have been unable to avoid the collision even if they were driving within the speed limit — no liability results. See 4 Fowler V. Harper, Fleming James, Jr., Oscar S. Gray, The Law of Torts § 20.5, at 165 (2d ed. 1986).
In a sense, the causal link requirement and the but for requirement are two different but related ways of asking whether a defendant’s actions were a substantial factor in causing the injury. Causal link says that, even if defendant’s wrong was a but for cause of the injury in a given case, no liability ensues unless defendant’s wrong increases the chances of such harm occurring in general. But for says even if what the defendant did greatly increased the risk of certain injuries occurring, unless it was a sine qua non of the specific harm that actually came about, no liability will be assessed.