This appeal requires us to explore the limits of a trial court's authority to exclude scientific evidence--in this instance, evidence of alleged cocaine use by the driver of a motor vehicle involved in a fatal accident and evidence of his ensuing impairment--under Daubert v. Merrell Dow Pharm., Inc., 509 U.S. 579, 113 S. Ct. 2786, 125 L. Ed. 2d 469 (1993). We conclude that the court below abused its discretion in excluding certain of this evidence and that the magnitude of the error necessitates a new trial.

On an afternoon in September of 1992, Julio Elvin Ruiz Cintrn (Ruiz) was driving a Toyota automobile westerly along a two-lane road in Puerto Rico. At what proved to be the critical moment, Ruiz left his lane to pass a slow-moving vehicle. Seconds later, his car collided with an oncoming eastbound tractor-trailer rig. Ruiz, his wife, son, and three other passengers (all minors) were killed. Ruiz's four-year-old daughter survived, but sustained permanent brain damage.

In due course, Ruiz's daughter, joined by other relatives of the various decedents (all of whom, at the time of suit, were citizens of mainland states), invoked diversity jurisdiction, 28 U.S.C. § 1332(a), and brought suit for damages in Puerto Rico's federal district court. They named as defendants the driver of the tractor-trailer unit, Juan Hernandez Rosario (Hernandez); his employer, Los Vaqueros de Transporte y Carga; the consignor, Pepsi Cola of Puerto Rico Bottling Company; and several insurers. The plaintiffs averred (1) that Hernandez needlessly accelerated his rig as the Toyota approached, thereby shortening the available time within which Ruiz could complete his passing maneuver and return to his own side of the road, and (2) that Hernandez refused to veer to the right to avoid the accident, despite having sufficient space and time to do so. The defendants denied the essential allegations of the complaint. They argued that Ruiz, and Ruiz alone, had caused the accident by recklessly initiating a passing maneuver in the face of obvious danger and placing his vehicle on the wrong side of the road. To bolster this thesis, the defendants sought to show that cocaine intoxication provoked Ruiz's recklessness.

The district court stymied the defendants' anticipated trial strategy by refusing to admit into evidence either the toxicology section of the autopsy report (which reflected the presence of cocaine and cocaine metabolites in Ruiz's bloodstream) or expert testimony regarding the significance of these findings. In the court's view, the proposed expert testimony failed to meet the standard of reliability required under Daubert. Having rejected the expert testimony, the court then excluded the toxicology results under Fed.R.Evid. 403, concluding that those results, standing alone and unexplained, were more prejudicial than probative.

In diversity cases, local law provides the substantive rules of decision. See Erie R.R. Co. v. Tompkins, 304 U.S. 64, 78, 58 S. Ct. 817, 82 L. Ed. 1188 (1938); Daigle v. Maine Med. Ctr., Inc., 14 F.3d 684, 689 (1st Cir.1994). Puerto Rico law recognizes comparative negligence principles. See P.R. Laws Ann. tit. 31, § 5141 (1990). The trial judge instructed accordingly, and the jury found both drivers negligent, assigning 59% of the fault to Ruiz and 41% to Hernandez. It then awarded damages totaling approximately $13,000,000 to the various plaintiffs. Due to Puerto Rico's combination of comparative negligence and joint and several liability rules, the defendants will be required to pay the full amount of these damages if the judgment becomes final.1

In the aftermath of the jury verdict, the defendants moved for judgment as a matter of law, Fed.R.Civ.P. 50, or, alternatively, for a new trial, Fed.R.Civ.P. 59. The district court rejected the defendants' plaints, including those that centered on the allegedly wrongful exclusion of the expert testimony and toxicology results. This appeal ensued.

The Daubert questions in this case are complex and implicate four interrelated pieces of evidence: (1) the toxicology results contained in the autopsy report; (2) the so-called "dosage" testimony, i.e., the expert opinions of a pharmacologist relating to the amount of drugs that Ruiz consumed and the time of their consumption, arrived at by interpolation from the toxicology results; (3) the so-called "impairment" testimony, i.e., the pharmacologist's expert opinions regarding the effects of cocaine on behavior; and (4) the so-called "causation" testimony, i.e., certain expert opinions of the defense's accident reconstructionist. After surveying the legal landscape, we discuss how these items came before the district court and how the court handled them. We then explicate the standard of review and proceed to test the correctness of the district court's rulings.

A. Daubert Revisited.

The Evidence Rules generally confine the testimony of a lay witness to matters about which he or she has personal knowledge, see Fed.R.Evid. 602, although such a witness may offer opinions that are "rationally based on [his or her] perception" and "helpful to a clear understanding of the witness' testimony or the determination of a fact in issue," Fed.R.Evid. 701. The Rules afford expert witnesses much more leeway. "If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or determine a fact in issue, a witness qualified as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise." Fed.R.Evid. 702. Despite its apparent breadth, this language does not give experts carte blanche, but, rather, envisions some regulation of expert testimony by trial judges. The Court's opinion in Daubert furnishes the principal source of guidance on the proper fulfillment of this gatekeeping role.

We start with an historical perspective. Prior to Daubert, courts and commentators regarded Frye v. United States, 293 F. 1013 (D.C.Cir.1923), as the watershed case on the admission of expert opinion testimony. Under Frye, the admissibility of an expert opinion or technique turned on its "general acceptance" vel non within the scientific community. Id. at 1014. Daubert tackled the question of whether the Frye standard survived the passage of the Federal Rules of Evidence (and, in particular, Rule 702) and answered that question in the negative, holding that Rule 702 displaced the Frye test. See Daubert, 509 U.S. at 587-89, 113 S. Ct. 2786.

The Daubert Court's interpretation of Rule 702, drawn from its text, requires the trial judge to evaluate an expert's proposed testimony for both reliability and relevance prior to admitting it. See id. at 589-95, 113 S. Ct. 2786. The requisite review for reliability includes consideration of several factors: the verifiability of the expert's theory or technique, the error rate inherent therein, whether the theory or technique has been published and/or subjected to peer review, and its level of acceptance within the scientific community. See id. at 593-95, 113 S. Ct. 2786. The Court reasoned that due investigation of such matters will ensure that proposed expert testimony imparts "scientific knowledge" rather than guesswork. Id. at 592, 113 S. Ct. 2786. Withal, the factors that the Court enumerated do not function as a "definitive checklist or test," but form the basis for a flexible inquiry into the overall reliability of a proffered expert's methodology. Id. at 593, 113 S. Ct. 2786.

Along with the reliability requirement, the Daubert Court imposed a special relevancy requirement. See id. at 591-92, 113 S. Ct. 2786. To be admissible, expert testimony must be relevant not only in the sense that all evidence must be relevant, see Fed.R.Evid. 402, but also in the incremental sense that the expert's proposed opinion, if admitted, likely would assist the trier of fact to understand or determine a fact in issue, see Daubert, 509 U.S. at 591-92, 113 S. Ct. 2786. In other words, Rule 702, as visualized through the Daubert prism, "requires a valid scientific connection to the pertinent inquiry as a precondition to admissibility." Id. at 592, 113 S. Ct. 2786.

In General Elec. Co. v. Joiner, 522 U.S. 136, 118 S. Ct. 512, 139 L. Ed. 2d 508 (1997), the Justices established the appropriate standard of appellate review for Daubert determinations, concluding that a reviewing tribunal should scrutinize a trial court's decision to allow or disallow the admission of expert testimony on Daubert grounds for abuse of discretion. See id. 118 S.Ct. at 517. Joiner also placed a gloss on Daubert 's insistence that trial courts focus on an expert's methodology, rather than his conclusions, in order to determine the reliability of his testimony. See Daubert, 509 U.S. at 595, 113 S. Ct. 2786. The Joiner Court moderated this position, acknowledging that

conclusions and methodology are not entirely distinct from one another. Trained experts commonly extrapolate from existing data. But nothing in either Daubert or the Federal Rules of Evidence requires a district court to admit opinion evidence which is connected to existing data only by the ipse dixit of the expert. A court may conclude that there is simply too great an analytical gap between the data and the opinion proffered.

Joiner, 118 S. Ct. at 519. Thus, while methodology remains the central focus of a Daubert inquiry, this focus need not completely pretermit judicial consideration of an expert's conclusions. Rather, trial judges may evaluate the data offered to support an expert's bottom-line opinions to determine if that data provides adequate support to mark the expert's testimony as reliable.2

Daubert and Joiner, though critically important, do not represent the sum total of available jurisprudential insights. Since Daubert hove into view, the courts of appeals have made significant contributions to an understanding of how to separate reliable from unreliable science and how to apply the intuitive idea of "fit"--as courts have come to call the special kind of relevance that Daubert demands--to live litigation scenarios. See, e.g., Baker v. Dalkon Shield Claimants Trust, 156 F.3d 248, 252-54 (1st Cir.1998); Daubert v. Merrell Dow Pharm., Inc., 43 F.3d 1311, 1316-22 (9th Cir.1995). Nonetheless, choreographing the Daubert pavane remains an exceedingly difficult task. Few federal judges are scientists, and none are trained in even a fraction of the many scientific fields in which experts may seek to testify. Moreover, even though Daubert and its progeny require trial judges to evaluate the level of support provided by complex scientific studies and experiments in myriad disciplines, reliability and relevance remain legal judgments. Trial judges cannot abdicate the responsibility for making those judgments by delegating them to the scientific community.

To complicate matters further, Daubert issues rarely arise in a vacuum, but, rather, frequently collide in practice with the requirements of other rules of evidence, especially Fed. R. Evid. 403.3 This phenomenon adds yet another dimension to the decisional calculus. See, e.g., Baker, 156 F.3d at 254.

B. The Challenged Evidence.

We turn now from the general to the particular, and canvass the evidence tendered below insofar as it bears on this appeal. Following Ruiz's death, the authorities ordered an autopsy. The autopsy report included toxicology results indicating that Ruiz's body contained 0.45 mcg/ml of cocaine and 0.15 mcg/ml of a cocaine metabolite (benzoylecgonine) in its blood, cocaine metabolites in its urine and vitreous humor, and cocaine in its nasal passages. The defendants proffered Dr. James O'Donnell, a well-credentialed pharmacologist, to comment upon the significance of these findings. Dr. O'Donnell was prepared to testify, based on the autopsy report, that in his opinion Ruiz had snorted 200 milligrams of cocaine within an hour prior to the accident. Dr. O'Donnell also proposed to testify that cocaine impairs senses and capabilities affecting driving, diminishes perception, and increases the willingness to take risks. The district court excluded not only Dr. O'Donnell's testimony, but also the toxicology results. And, though the court permitted the defendants' accident reconstruction expert, Eric Cintrn, to testify, the court apparently precluded counsel from eliciting his opinion that Ruiz would not have initiated the fateful passing maneuver had it not been for his cocaine intoxication.4

C. Proceedings Below.

The evidentiary issues in this case came to the fore on February 28, 1997, when the plaintiffs filed motions in limine seeking the exclusion of both Dr. O'Donnell's testimony and the toxicology results. The plaintiffs argued that levels of cocaine and metabolites in the bloodstream could not be correlated to initial dosage or levels of impairment with the degree of certainty required for scientific evidence, and they produced several articles to this effect. They sought exclusion of the toxicology results mainly on the theory that the chain of custody of the samples of bodily fluids had been compromised. The defendants promptly submitted an opposition. They proffered Dr. O'Donnell's report, the report of the plaintiffs' expert toxicologist (suggesting that he used a methodology similar to that employed by Dr. O'Donnell), and portions of plaintiffs' expert's deposition. Based on these materials, they argued that Dr. O'Donnell's methodology was sufficiently reliable to withstand Daubert scrutiny. As to the toxicology results, the defendants contended that the plaintiffs' argument was factually wrong, and that, in all events, any gaps in the chain of custody went to the weight of the evidence, not to its admissibility. See, e.g., United States v. Ladd, 885 F.2d 954, 956 (1st Cir.1989).

The district court held a pretrial conference on March 5, 1997. At the court's request, defense counsel agreed to furnish copies of the 18 articles cited by Dr. O'Donnell in his report, and submitted 14 of them within the next few days. Without waiting for the remainder, the district court ruled on March 10. It concluded that there was "no scientific basis" for Dr. O'Donnell's dosage opinion (i.e., his opinion that Ruiz had snorted at least 200 mg of powdered cocaine within an hour before the accident) and that Dr. O'Donnell's impairment opinion (i.e., his opinion that cocaine ingestion had impaired Ruiz's driving ability) likewise failed to pass the Daubert reliability screen. To close the circle, the court barred introduction of the toxicology results "in light of" the exclusion of Dr. O'Donnell's testimony.

Trial began the next day. On March 26, the plaintiffs moved in limine to exclude the testimony of Eric Cintrn. Noting a statement in Cintrn's report to the effect that Ruiz "appears to have been driving under the influence of drugs as it is evidenced by the toxicological report," and his conclusion that "this may explain the reason for [the] ... passing maneuver," the plaintiffs argued for exclusion both because the trial court had banished the toxicology results and because Cintrn was unqualified to opine on the relationship between cocaine and driving ability. Although the district court never formally ruled on this motion, Cintrn was not asked his opinion on these matters when he testified, and the parties treat this omission as flowing from the district court's prohibition on any mention of the toxicology results.

On March 31--the same day that the plaintiffs completed their case in chief--the defendants filed a motion to admit Dr. O'Donnell's testimony and attached to it copies of numerous additional articles and portions of learned treatises regarding cocaine's absorption, excretion, detection in bodily fluids, and effects on the senses and behavior. Judge Acosta denied the motion the following day, terming it "untimely." Other motions for reconsideration, made at divers points during the presentation of the defense's case, met a similar fate.

D. Standard of Review.

Nobody questions that testimony regarding dosage--what quantity of drugs Ruiz consumed, and when--and the effect of Ruiz's cocaine consumption on his driving ability would be relevant as long as it could be reliably determined that Ruiz ingested an appreciable amount of cocaine shortly before the accident and that such dosage probably would impair one's ability to drive. Thus, although the Daubert standard requires that expert testimony be both reliable and relevant as a precondition to admissibility, the reliability inquiry is central in this case.

The district court concluded that Dr. O'Donnell's testimony on both dosage and impairment lacked scientific reliability and precluded the testimony on that basis. Forbidding the toxicology results and Cintrn's opinion on causation followed directly from this exclusionary ruling. We review all these determinations for abuse of discretion. See Joiner, 118 S. Ct. at 517.

While this standard of review ordinarily is "not appellant-friendly," Lussier v. Runyon, 50 F.3d 1103, 1111 (1st Cir.1995), it does not render trial court decisions impervious to scrutiny. On abuse-of-discretion review, we will reverse a trial court's decision if we determine that the judge "committed a meaningful error in judgment." Anderson v. Cryovac, Inc., 862 F.2d 910, 923 (1st Cir.1988); see also Foster v. Mydas Assocs., Inc., 943 F.2d 139, 143 (1st Cir.1991) (explaining that an abuse of discretion occurs "when a material factor deserving significant weight is ignored, when an improper factor is relied upon, or when all proper and no improper factors are assessed, but the court makes a serious mistake in weighing them") (quoting Independent Oil & Chem. Workers of Quincy, Inc. v. Procter & Gamble Mfg. Co., 864 F.2d 927, 929 (1st Cir.1988)).

E. Analysis.

To determine whether the nisi prius court committed "a meaningful error in judgment" here, we assay its evaluation of each contested piece of evidence under the Daubert standard.

1. The Dosage Testimony. We first consider the exclusion of Dr. O'Donnell's testimony concerning the approximate time of Ruiz's cocaine consumption and the amount of cocaine actually ingested by him. Based on autopsy data--cocaine and cocaine metabolites found in Ruiz's blood, urine, nasal passages, and vitreous humor--and half-life values for the substances involved,5 Dr. O'Donnell applied a mathematical formula and concluded "with reasonable pharmacological certainty" that Ruiz " 'snorted' at least 200-mg of cocaine powder into his nose thirty (30) to sixty (60) minutes prior to the accident." Although Dr. O'Donnell's report cites numerous scientific writings in support of the methodology underlying this proposition, the lower court found none of these sources adequate to imbue the proffered opinions with the patina of reliability required by Daubert.

We have read the articles mentioned by Dr. O'Donnell and supplied to the district court.6 Several of them discuss a halflife for cocaine or its metabolites and/or the times after ingestion at which cocaine and its metabolites reach peak concentrations in the body. One of these sources is a standard medical textbook. See Matthew J. Ellenhorn & Donald G. Barceloux, Medical Toxicology 648-49 (1988) (placing half-life in blood plasma for cocaine administered intransally at 1.3 hours and citing the average half-life for such cocaine in urine at 75 minutes). Another is a published article in a prestigious, peer-reviewed medical journal. See Peter M. Marzuk, et al., Fatal Injuries After Cocaine Use As a Leading Cause of Death Among Young Adults in New York City, 332 New Eng. J. Med. 1753, 1754 (1995) (using half-life of cocaine metabolites in urine to determine use of cocaine by victims of fatal car accidents). The publication of these pieces and their exposure to peer review serve as independent indicia of the reliability of the half-life technique. By the same token, publication and peer review also demonstrate a measure of acceptance of the methodology within the scientific community.

Secondary sources cited by Dr. O'Donnell lack publication and peer review, see, e.g., Daniel S. Isenschmid, Cocaine (Plasma Concentrations of Cocaine Metabolites in Humans ) 11-12 (collecting and reporting studies finding half-lives for cocaine metabolites in urine and peak levels of cocaine in plasma after intranasal administration of cocaine), but this circumstance does not make such sources per se unacceptable, see Daubert, 509 U.S. at 593, 113 S. Ct. 2786 (explaining that neither publication nor peer review is "a sine qua non of admissibility"). Under ordinary circumstances, an unpublished, unreviewed work, standing alone, probably would be insufficient to demonstrate the reliability of a scientific technique. But when such an article makes the same point as published, peer-reviewed pieces, it tends to strengthen the assessment of reliability.

Other works referenced by Dr. O'Donnell detail controlled studies correlating dosages with levels of cocaine (and its metabolites) remaining in the body after certain periods of time. See, e.g., Randall C. Baselt & Robert H. Cravey, Disposition of Toxic Drugs and Chemicals in Man 208-09 (1989) (collecting studies); H.E. Hamilton, et al., Cocaine and Benzoylecgonine Excretion in Humans, 22 J. Forensic Sci. 697, 698-706 (1987) (reporting results of a study that administered cocaine intranasally to healthy subjects and tested urine for levels of cocaine and its metabolites after 1, 2, 4, 8, 12, 24, 48, 72, 120, 144, and 168 hours). These manuscripts further confirm the reliability of Dr. O'Donnell's approach by providing information that can be used to test the accuracy of the technique upon which he relied. And, finally, the fact that the plaintiffs' expert employed essentially the same technique furnishes added validation.

To be sure, the scientific literature does not make out an open-and-shut case. In defense of the trial court's ruling, the plaintiffs point to statements (some of which appear in works referenced by Dr. O'Donnell) that the half-life of cocaine varies among individuals due to many factors, and that cocaine metabolites can be found in the body days after ingestion. See, e.g., Ellenhorn & Barceloux, supra at 649 (stating that the half-life of cocaine in urine "varies significantly between individuals"); Hamilton, et al., supra, at 703 (finding cocaine metabolites in subjects' urine up to 120 hours after insufflation of cocaine). The plaintiffs posit that these statements indicate the inherent unreliability of Dr. O'Donnell's dosage opinion. If half-life varies, they ruminate, it cannot be used to determine initial dosage with any accuracy--and the discovery of metabolites offers no more information because these may remain in the body long after the drug's effects on behavior have subsided. Thus, the plaintiffs asseverate, Daubert forbids the admission of Dr. O'Donnell's testimony.

We think that the plaintiffs (and the district court) set the bar too high. Although the statements that they assemble cast doubt on Dr. O'Donnell's position--for example, those statements suggest that the half-life technique for calculating dosage has an uncertain rate of error--no single factor disposes of a reliability inquiry. See Daubert, 509 U.S. at 592-95, 113 S. Ct. 2786. Dr. O'Donnell's technique has been subjected to, and survived, the rigors of testing, publication, and peer review, and it appears to have won significant (if not universal) acceptance within the scientific community. Daubert does not require that a party who proffers expert testimony carry the burden of proving to the judge that the expert's assessment of the situation is correct. As long as an expert's scientific testimony rests upon "good grounds, based on what is known," Daubert, 509 U.S. at 590, 113 S. Ct. 2786 (internal quotation marks omitted), it should be tested by the adversary process--competing expert testimony and active cross-examination--rather than excluded from jurors' scrutiny for fear that they will not grasp its complexities or satisfactorily weigh its inadequacies, see id. at 596, 113 S. Ct. 2786. In short, Daubert neither requires nor empowers trial courts to determine which of several competing scientific theories has the best provenance. It demands only that the proponent of the evidence show that the expert's conclusion has been arrived at in a scientifically sound and methodologically reliable fashion. See Kannankeril v. Terminix Int'l, Inc., 128 F.3d 802, 806 (3d Cir.1997); In re Paoli R.R. Yard PCB Litig., 35 F.3d 717, 744 (3d Cir.1994).

On balance, we find that Dr. O'Donnell's dosage opinion, incorporating a range of time in which he believed Ruiz took the cocaine, satisfies this standard. The opinion was premised on an accepted technique, embodied a methodology that has significant support in the relevant universe of scientific literature, and was expressed to a reasonable degree of pharmacological certainty. While the literature does not irrefutably prove the accuracy of Dr. O'Donnell's dosage conclusions, it furnishes a sufficient underpinning for those conclusions to forfend preclusion of his testimony as unreliable. Thus, the district court's refusal to entertain Dr. O'Donnell's dosage opinion constituted an abuse of discretion. See Baker, 156 F.3d at 252-54; see also Ed Peters Jewelry Co. v. C & J Jewelry Co., 124 F.3d 252, 258-61 (1st Cir.1997) (finding, prior to Joiner, exclusion of expert testimony erroneous under either abuse of discretion or a more stringent standard of review).

2. The Impairment Testimony. Although Dr. O'Donnell's opinion on dosage contained sufficient indicia of reliability to warrant its admission into evidence, the defendants needed other expert evidence to flesh out their theory as to how the accident occurred. In particular, they had to show a correlation between cocaine use in the dosage suggested by Dr. O'Donnell's opinion and an impairment affecting the cocaine user's fitness to drive. Dr. O'Donnell was ready to supply the missing link: he opined that cocaine intoxication resulting from a dose such as Ruiz insufflated results in impairments of perception, reflexes, reaction time, and judgment, and that such a degree of intoxication increases one's "sense of mastery" and thus promotes risk taking. Virtually all of these sequelae would adversely affect the ability to drive safely.

The plaintiffs urged the trial court to hold that Dr. O'Donnell's impairment testimony failed Daubert 's reliability prong because science cannot correlate particular impairments to cocaine concentrations within the body. See, e.g., Peter M. Marzuk, et al., Prevalence of Recent Cocaine Use Among Motor Vehicle Fatalities in New York City, 253 J. Am. Med. Ass'n 250, 255 (1990). In the plaintiffs' view, only an immediate neurological examination could have provided sufficiently reliable evidence as to whether Ruiz suffered from cocaine intoxication at the time of the accident (and if so, to what extent)--and no such examination was performed.

The court granted the plaintiffs' motion to exclude the impairment testimony. We quote the core of the court's reasoning, as explicated in its subsequent denial of the defendants' motion for a new trial:

All the scientific literature submitted to the Court on this issue unanimously concluded that unlike alcohol, a correlation between a particular amount of cocaine in the system and the degree of impairment produced has not been scientifically determined because the effects of this substance var[y] from one individual to another. Thus, from the scientific evidence reviewed by the Court it was concluded that even though it is generally accepted that cocaine causes impairment in an individual [it] cannot be scientifically deduced from the amount found in his or her system because persons metabolize cocaine differently.

We believe that the court's rationale conflated the dosage and impairment issues. The court agreed that "cocaine causes impairment," but rejected Dr. O'Donnell's proposed testimony on this point because of its distrust of the witness's dosage testimony. By relying so heavily on an improper factor in the decisional calculus--the dosage testimony, as we have said, was sufficiently reliable to satisfy Daubert--the district court abused its discretion. See Foster, 943 F.2d at 143; Independent Oil Workers, 864 F.2d at 929.

To compound this error, the court applied a standard of scientific certainty to the impairment testimony beyond that which Daubert envisions. The court imposed a threshold requirement that science be able to declare that a precise quantity of cocaine in the bloodstream produces an equally precise degree of impairment. This requirement solicits a level of assurance that science realistically cannot achieve and that Daubert does not demand. See Daubert, 509 U.S. at 590, 113 S. Ct. 2786 (commenting that "arguably, there are no certainties in science"). The adoption of such a standard impermissibly changes the trial judge's role under Daubert from that of gatekeeper to that of armed guard. That mistaken application of the law likewise constitutes an abuse of discretion. See United States v. Snyder, 136 F.3d 65, 67 (1st Cir.1998) (holding that a per se abuse of discretion occurs when a district court commits an error of law).

3. The Remaining Evidence. Having decided to ban Dr. O'Donnell's proffered testimony, the district court held that the prejudicial effect of admitting the toxicology results would substantially outweigh their probative value and excluded this evidence under Rule 403. This is perfectly understandable: to allow the jury to be told that traces of cocaine were found in Ruiz's body without any accompanying explanation of the meaning of the test results or of cocaine's capacity to impair driving skills would sow the seeds for an horrific harvest. It is evident, however, that the lower court's rulings regarding the admissibility of Dr. O'Donnell's proffered testimony influenced the calibration of the Rule 403 balance. The error inherent in those rulings requires vacation of this aspect of the district court's order vis-a-vis the toxicology results.7

The district court's prohibition of Cintrn's causation opinion also must be reconsidered. Cintrn's proposed testimony as to cocaine's role in the accident was based on other evidence (e.g., the toxicology results) that the district court erroneously excluded. This revue nullifies the most obvious reason for precluding Cintrn from testifying anent causation. As with the toxicology results, this ruling will have to be revisited by the district court on remand.F. Effect of the Errors.

Our odyssey is not yet concluded. The plaintiffs contend that even if the trial court erred in excluding evidence anent Ruiz's use of cocaine, the mistakes are harmless and do not require a new trial. See Fed.R.Civ.P. 61; Fed.R.Evid. 103.

The plaintiffs' argument is not without some force. Puerto Rico is a comparative negligence jurisdiction that imposes joint and several liability on joint tortfeasors. This doctrinal combination means that a driver whose negligence is found to have contributed to causing an accident may be solely responsible for compensating the victims even though the other driver involved in the accident was more negligent. See generally Ramos Acosta, 116 P.R. Offic. Trans. at 81-82.

In this case, the district court properly directed the jury to make findings anent comparative negligence, and the jury assigned fault to both drivers (59% to Ruiz, 41% to Hernandez). The plaintiffs maintain that the evidence of cocaine use was relevant only to Ruiz's negligence and that its prohibitation did not undermine the jury's finding that Hernandez's negligence contributed significantly to the accident's occurrence.8 Since Hernandez's negligence subjects the defendants to liability for the full amount of the verdict, see supra note 1, the plaintiffs hypothesize that the improper exclusion of the cocaine evidence does not require a new trial. We reject this hypothesis.

The draconian potential of the Puerto Rico rules is mitigated by the theories of efficient cause and absorption. Under the principle of efficient cause, when one party is the "sole, and efficient cause of the damage," another's negligence during the accident, if it did not cause it, does not subject him to liability. Toro Lugo v. Ortiz Martinez, 113 P.R. Offic. Trans. 73, 75, 113 P.R. Dec. 56, 56, 1982 WL 210594 (1982). Under the absorption theory, if one tortfeasor is only slightly responsible, the overwhelming negligence of the other tortfeasor "absorbs" the minimal negligence of the former and the latter bears all liability. See id.; see also Santiago v. Becton Dickinson & Co., 571 F. Supp. 904, 911-14 (D.P.R.1983) (comparing efficient cause and absorption principles).

These theories have particular pertinence here. Ruiz's alleged cocaine intoxication formed the foundation not only for the defendants' insistence that Ruiz was guilty of negligence, but also for their insistence that Hernandez was not negligent. With the benefit of the cocaine evidence, the jurors may well have balanced the proofs of negligence quite differently. After all, they found Ruiz 59% negligent even without the damaging evidence suggesting that his judgment and driving ability may have been impaired by cocaine at the time of the crash. The additional evidence, if ultimately shown to be admissible, easily could lead a rational jury to find Ruiz's negligence to have been so great as to overwhelm Hernandez's negligence.

An erroneous evidentiary ruling requires vacation of a jury verdict if the ruling excludes evidence and "the exclusion results in actual prejudice because it had a substantial and injurious effect or influence in determining the jury's verdict." United States v. Shay, 57 F.3d 126, 134 (1st Cir.1995) (citation and internal quotation marks omitted). In order to determine whether a particular ruling had a sufficiently pernicious effect on a verdict, a reviewing court must resist the temptation to deal in abstractions, and must mull the ruling in context, giving due weight to the totality of the relevant circumstances. See Nieves-Villanueva v. Soto-Rivera, 133 F.3d 92, 102 (1st Cir.1997). To sustain the verdict, the reviewing court must be able to say with a fair degree of assurance that the erroneous ruling did not substantially sway the jury. See Ladd, 885 F.2d at 957 (citing Kotteakos v. United States, 328 U.S. 750, 765, 66 S. Ct. 1239, 90 L. Ed. 1557 (1946)). We can muster no such assurance here.

We reach this conclusion despite the somewhat awkward procedural posture in which this appeal arises. In the prototypical "exclusion of evidence" case, the court of appeals assesses the effect that a piece of evidence, wrongly withheld from the jury, likely would have had if introduced. See, e.g., Espeaignnette v. Gene Tierney Co., 43 F.3d 1, 5-9 (1st Cir.1994). This case is different. Even with the misplaced Daubert obstacle removed, the dosage testimony is not necessarily admissible. Rather, its admissibility depends to some extent upon whether certain other evidence, specifically, the toxicology results and the impairment testimony, prove to be admissible. Without the former (to establish the basis for Dr. O'Donnell's dosage opinion) and the latter (to explain the probable effects of Ruiz's cocaine use on his driving), the dosage testimony, though deemed reliable in the Daubert sense, might not be admissible in the long run. Yet, when a trial court erroneously excludes evidence, and the exclusion meets the standard criteria of harmfulness, the harm is not cured by a mere possibility that other appropriate grounds for exclusion of the same evidence may later be found to exist. The question is one of degree and the choice of remedies (including whether to require a new trial or merely remand for further findings) is ours. See 28 U.S.C. § 2106.

In this instance, we think that the defendants' chances of succeeding in their effort to introduce the cocaine-related evidence are promising enough that the improper exclusion of the dosage testimony, coupled with the errors regarding the court's consideration of the impairment testimony, can be said to have materially curtailed the defendants' opportunity to present their theory of the case to the jury. This, in turn, worked a substantial and injurious effect on the jury's ability to evaluate liability. Taking into account all aspects of the situation, we are persuaded that vacation of the judgment, rather than a remand for further findings, is the fairest course. A new trial will allow a judge appropriately to ascertain the admissibility of expert testimony and a jury armed with all reliable and relevant evidence to weigh issues of comparative fault on a scale that is in balance. To this end, we reject the plaintiffs' claim that any error committed by the trial court was benign.

We need go no further.9 By excluding Dr. O'Donnell's dosage testimony, the district court exceeded the scope of its discretion under Daubert. This fundamental error infected certain other evidentiary rulings. On the facts of this case, there is too great a risk that these rulings, in cumulation, had a substantial and injurious influence upon the jury's determinations. Hence, we reverse the district court's Daubert ruling, vacate the judgment below, and order a new trial. We emphasize that, in doing so, we hold only that the dosage testimony is sufficiently reliable to be admissible. We do not pass upon the ultimate admissibility of that evidence (which, as we have said, depends upon the admissibility of the toxicology results, see supra note 7, and the admissibility of the correlative impairment testimony). By like token, we do not pass upon the ultimate admissibility of either the impairment or causation testimony. That evidence must be reexamined within the limits dictated by Daubert and its progeny.

Reversed and remanded for a new trial.