The issue presented by this appeal is how a court should determine the admissibility of expert opinion testimony. At the summary judgment stage of this case, plaintiff attempted to establish medical causation of a toxic tort through the testimony of a single expert witness. The district court held that the basis of the expert's opinion was insufficiently reliable and, in the alternative, that the expert's testimony would have been more prejudicial than probative. With the expert's testimony ruled inadmissible, plaintiff was left without proof of causation. The district court entered summary judgment for the defendants. We affirm.

Christophersen died in March of 1986 as a result of a rare, small-cell form of cancer that originated in his colon and metastasized to his liver. During the fourteen years preceding his death, Christophersen worked for Marathon at its plant in Waco, Texas. At that plant, Marathon produces nickel/cadmium batteries. Christophersen never was directly involved in the production of these batteries. The record, however, indicates that over a number of years Christophersen's job duties required him to visit the area of the plant in which the batteries were manufactured. During these visits, Christophersen was allegedly exposed to fumes resulting from the manufacturing process. Plaintiffs, Christophersen's surviving spouse and child, contend that these fumes contained particles of nickel and cadmium and that Christophersen's exposure to these heavy metals caused the cancer that resulted in his death.

Plaintiffs brought suit pursuant to the Texas Wrongful Death and Survival Statute, Tex.Civ.Prac. & Rem.Code Ann. Secs. 71.001-.031 (Vernon 1986), against Marathon and a number of companies that supplied Marathon with chemicals and other materials used in the manufacture of the nickel/cadmium batteries. Plaintiffs' complaint alleged that the products used in the production of the batteries were defectively designed, manufactured, and marketed, and were the producing causes of the cancer that resulted in Christophersen's death. The complaint also alleged that Marathon was aware of the dangerous nature of the chemicals and products and failed to provide Christophersen a safe place to work or to warn him of the dangerous conditions that existed at the plant. Marathon moved for summary judgment. The district court determined that plaintiffs failed to state a design defect or manufacturing defect claim and granted the defendants' motions for summary judgment. Plaintiffs do not appeal this ruling.

The district court also granted Marathon's motion for summary judgment on the marketing defect claim because the plaintiffs did not present sufficient evidence of causation. In reaching this conclusion, the court focused on the affidavit of the plaintiffs' expert witness, Dr. Miller, who concluded that Christophersen's exposure to nickel and cadmium at Marathon caused the cancer that resulted in his death. The district court undertook an in-depth review of the basis for Dr. Miller's conclusion and determined that his opinion should be excluded. On appeal, a panel of this court reversed, holding that Dr. Miller's opinion was not so fundamentally unreliable that the jury should not consider it. Defendants timely petitioned this court for rehearing en banc.

A trial court's ruling regarding admissibility of expert testimony is protected by an ambit of discretion and must be sustained unless manifestly erroneous.1 At the same time, we accord "proper deference to the jury's role as the arbiter of disputes between conflicting opinions. As a general rule, questions relating to the bases and sources of an expert's opinion affect the weight to be assigned that opinion rather than its admissibility and should be left for the jury's consideration." Viterbo v. Dow Chem., 826 F.2d 420, 422 (5th Cir.1987); see also Dixon v. International Harvester, 754 F.2d 573, 580 (5th Cir.1985).

This is an appeal from a grant of summary judgment. Although we review grants of summary judgment de novo, that is, under the same Rule 56 standards as are used by the district court, e.g., Medlin v. Palmer, 874 F.2d 1085, 1089 (5th Cir.1989) (applying the Celotex and Rule 56 summary judgment standards de novo) (citing United States Steel Corp. v. Darby, 516 F.2d 961 (5th Cir.1975)), in Rule 56 proceedings we still apply the manifest-error standard of review to the trial court's evidentiary rulings, Lavespere v. Niagara Mach. & Tool Works, 910 F.2d 167, 175-76 (5th Cir.1990); Slaughter v. Southern Talc Co., 919 F.2d 304, 306-07 (5th Cir.1990); Washington v. Armstrong World Indus., 839 F.2d 1121, 1123 (5th Cir.1988); Viterbo, 826 F.2d at 422.2 Thus an appeal of a summary judgment presenting evidentiary issues raises two levels of inquiry. At the first level, we review the trial court's evidentiary rulings, which define the summary judgment record, and we give these rulings their due deference. At the second level, with the record defined, we review the trial court's summary judgment decision de novo. When the contested evidence is essential to the cause of action and the trial court has excluded the evidence, we may decide the appeal at the first level solely on the basis of the soundness of the evidentiary ruling. For if we uphold the exclusion of essential evidence, the second-level inquiry becomes academic. E.g., Viterbo, 826 F.2d at 422 (stating that appropriateness of summary judgment depends solely upon whether the district court erred in excluding causation testimony of plaintiff's expert). This is such a case. Without Dr. Miller's testimony, Christophersen cannot prove that any exposure to nickel/cadmium caused the colon cancer. In sum, we ask in this appeal whether excluding Dr. Miller's opinion was manifestly erroneous.

The Federal Rules of Evidence, combined with Frye v. United States, 293 F. 1013 (D.C.Cir.1923), provide a framework for trial judges struggling with proffered expert testimony. The signals are not neatly cabined categories, and we

disentangle them only to accent the independent significance of each.

(1) Whether the witness is qualified to express an expert opinion, Fed.R.Evid. 702;

(2) whether the facts upon which the expert relies are the same type as are relied upon by other experts in the field, Fed.R.Evid. 703;

(3) whether in reaching his conclusion the expert used a well-founded methodology, Frye; and

(4) assuming the expert's testimony has passed Rules 702 and 703, and the Frye test, whether under Fed.R.Evid. 403 the testimony's potential for unfair prejudice substantially outweighs its probative value.

These four signals or inquiries introduce no new concepts to our jurisprudence. They are only guideposts drawn from the Federal Rules of Evidence and our cases. We list these inquiries, but in doing so we do not intend that they be applied mechanically. At the same time, they often will naturally lend themselves to sequential application. The reality is that trials are too varied for fixed molds; we construct none today.

The first three steps are best understood as threshold requirements that all expert testimony must meet before being deemed admissible. Cf. Gideon v. Johns-Manville Sales Corp., 761 F.2d 1129, 1135 (5th Cir.1985) (construing Rule 702's qualifications requirement as a threshold inquiry); Slaughter, 919 F.2d at 306-07 (construing Rule 703's factual basis requirement as a threshold inquiry). Rule 403, on the other hand, provides an overlay--a final mechanism for screening out otherwise admissible testimony whose potential for prejudice substantially outweighs its probative value.3 Cf. 22 C. Wright & K. Graham, Federal Practice & Procedure, Sec. 5213 at 258-59 (1978).

The first question concerns the expert's qualifications: Is the witness--because of his specialized knowledge, skill, experience, training, or education in the relevant field--qualified to express an expert opinion on the topic at issue? Fed.R.Evid. 702. The Advisory Committee Note accompanying Rule 702 reads the broad language of the rule to permit expert testimony not only by experts carrying formal credentials such as university degrees and professional memberships but also by so-called skilled witnesses, whose experiences permit them to testify with authority on a given topic. The areas of inquiry that expert testimony may address are similarly broad, including scientific and technical questions as well as any other area of specialized knowledge. An expert may testify in his area of expertise "in the form of an opinion or otherwise." This much is rote. The more subtle problem, and our caveat, is that the inquiry into the qualifications of an expert should not be a substitute for scrutinizing an expert's reasoning or methodology. At this stage, the only question for the trial court is whether the expert is generally qualified to render an opinion on the question in issue.

Second, if the expert is qualified, are the facts and data that serve as a basis for the expert's opinion the same type of facts as other experts in the same field reasonably rely upon in forming their opinions? Fed.R.Evid. 703.4 While testimony based on the personal observations of the expert is preferable, neither the rules nor our cases have insisted on personal examinations.5 The reports and statements of others such as doctors, nurses, or medical personnel, while not as valuable as testimony based on the expert's own observations, can provide a reliable basis for the expert's opinion, at least when reliance on such sources is the custom of the discipline. At the same time, a common-sense skepticism may be warranted when an expert's factual basis is derived, not from treatment or observation, but from subjective information obtained from counsel or client in preparation for trial.6 But such skepticism should not necessarily lead us to exclude the expert's opinion. So long as the facts upon which the expert bases his opinion are those "perceived or made known to the expert at or before the hearing" and are "of a type reasonably relied upon by experts in the particular field,"7 we should proceed to evaluate the expert's methodology. Fed.R.Evid. 703.

The third factor is the Frye test. We ask whether in reaching his conclusion, the witness used a well-founded methodology or mode of reasoning, one "sufficiently established to have gained general acceptance in the particular field in which it belongs." Frye, 293 F. at 1014; see also Osburn v. Anchor Laboratories, Inc., 825 F.2d 908 (5th Cir.1987), cert. denied, 485 U.S. 1009, 108 S. Ct. 1476, 99 L. Ed. 2d 705 (1988). As long as the expert's methodology is well founded, the nature of the expert's conclusion is generally irrelevant, even if it is controversial or unique. See Peteet v. Dow Chemical Co., 868 F.2d 1428 at 1433 (5th Cir.1989); Osburn, 825 F.2d at 915.8 In fact, in Osburn the plaintiff's and the defendant's experts relied on essentially the same diagnostic methodologies but drew opposite conclusions from the available information. We did not attempt to determine which expert's conclusion was more in line with the consensus in the scientific community. Instead we stated, "a jury must be allowed to make credibility determinations and weigh the conflicting evidence in order to decide the likely truth of a matter not itself initially resolvable by common knowledge or lay reasoning." Id. at 916. "An expert's opinion need not be generally accepted in the scientific community before it can be sufficiently reliable and probative in support of a jury finding." Osburn, 825 F.2d at 915 (emphasis added).9

Finally, Rule 403 serves a general screening function for otherwise admissible evidence. Assuming the witness's testimony is relevant and the witness is qualified, and assuming his testimony has the appropriate factual basis and his methodology is well founded, the only remaining inquiry is the balancing authorized in Rule 403: "Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence." In the 403 balancing inquiry, courts may consider issues that were appropriately excluded from the three preceding threshold inquiries, which are narrower in scope. If, taking all considerations into account, a court finds that the potential for prejudice substantially outweighs the probative value, the court may opt to exclude the testimony.10 Unlike Rules 702 and 703, which define admissibility thresholds for expert-opinion evidence, Rule 403 "creates a power to exclude otherwise admissible evidence; it gives no discretion to admit evidence that is subject to exclusion under some other rule." 22 C. Wright & K. Graham, Federal Practice & Procedure, Sec. 5213 at 258-59 (1978). We caution, however, that clearing the hurdles of Article VII of the Federal Rules of Evidence will not alone ensure approval under Rule 403. An application of these principles to this case follows.

Although the district court did not conclude that Dr. Miller was unqualified to testify under Rule 702, it did question his qualifications:

Dr. Miller is not an expert in either oncology or pathology. Miller's opinion as to the cause of Christophersen's death was formed without consultation with oncologists or other cancer specialists. Dr. Miller's experience with cancer occurred during his residency when he assisted in a study of the immune system as affected by smoking and asbestos. Dr. Miller does not routinely treat cancer patients, nor has he ever treated a patient with a colon cancer of the type that affected Christophersen.

(citations omitted).

The trial court's concerns are understandable. As we stated in Peteet, "district judges and appellate courts must carefully review an expert's testimony to ensure that the expert has the necessary qualifications and a sufficient basis for his opinion." 868 F.2d at 1431; Lavespere, 910 F.2d at 176 (citing Peteet ); see also In Re Air Crash Disaster at New Orleans, Louisiana, 795 F.2d 1230 at 1234-35 (5th Cir.1986) ("Trial judges must be sensitive to the qualifications of persons claiming to be an expert."). We caution, however, that although credentials can be significant, they alone are not necessarily determinative. The questions, for example, do not stop if the expert has an M.D. degree. That alone is not enough to qualify him to give an opinion on every conceivable medical question. This is because the inquiry must be into actual qualification--sufficient to assist the trier of fact. Viterbo, 826 F.2d at 422. The trial judge here rightly scrutinized Dr. Miller's lack of specialized experience and knowledge. The district court did not base its disallowance of Dr. Miller's testimony on this issue, however. It preferred to rely on Rule 703 and the witnesses' methodology. We, therefore, move to those issues.

Dr. Miller premised his opinion that Marathon caused Christophersen's cancer on his belief that Christophersen had approximately a twenty-year history of "extensive exposure to nickel and cadmium fumes in the work place." The district court, pursuant to Rule 703, analyzed the underlying "facts and data" of Dr. Miller's opinion to determine whether it was based on the types of facts reasonably relied upon by experts in the field.11 Dr. Miller testified at his deposition that the level and duration of the patient's exposure are important considerations when evaluating the effect of exposure to a toxic substance.

The district court found that virtually all of the factual data concerning Christophersen's exposure to nickel and cadmium came from the affidavit of a Marathon employee named Edgar Manoliu (Manoliu), who described the fumes and Christophersen's exposure to them. The district court criticized this affidavit, however, as being inaccurate and incomplete. The Manoliu affidavit appears to have over-estimated the number of times per week Christophersen visited the manufacturing area, as well as the average length of time he would remain there on each visit. The affidavit was also devoid of any information about the type of fumes to which Christophersen was exposed or the type of fumes generated by the battery manufacturing process.

We find particularly telling Manoliu's admission in his deposition that he did not know the chemical composition of the fumes nor the mix of chemicals in the impregnation and soak tanks.

Nor was Dr. Miller informed as to the physical facilities at the Marathon plant, including the size of the plant or the impregnation and soak area, or the ventilation available in these areas or in Christophersen's office. In addition, Dr. Miller did not always rely upon the accurate data that were contained in the affidavit. For example, the affidavit correctly indicated that Christophersen worked for fourteen years at the Waco plant before his death. Dr. Miller, however, based his opinion upon the assumption that Christophersen worked in the plant for twenty years. Thus Dr. Miller over-estimated the duration of Christophersen's exposure by approximately fifty percent. Accordingly, accurate dosage and exposure information was not used by Dr. Miller.

Thus the court validly called into question the facts and data relied upon by Dr. Miller in forming his opinion. Slaughter v. Southern Talc Co., 919 F.2d 304 (5th Cir.1990), reiterated the requirements under Rule 703:

As a general rule, questions regarding the scientific bases of an expert's opinion "affect the weight to be assigned that opinion rather than its admissibility and should be left for the jury's consideration." Viterbo v. Dow Chemical Co., 826 F.2d 420, 422 (5th Cir.1987). See also, Dixon v. International Harvester Co., 754 F.2d 573, 580 (5th Cir.1985). However, this general rule yields when "the source upon which an expert's opinion relies is of such little weight ... that [the] testimony would not actually assist the jury in arriving at an intelligent and sound verdict." Viterbo, 826 F.2d at 422. In such case, this court requires courts to examine the reliability of an expert's sources to determine whether they satisfy the threshold established by the rule. See Soden v. Freightliner Corp., 714 F.2d 498, 502 (5th Cir.1983).

Id. at 306-07.

Plaintiffs do not contest the district court's findings as to the deficiencies in the Manoliu affidavit. Rather, they argue that Dr. Miller stated in his opinion that dosage was less important when determining individual causation. Plaintiffs accordingly argue that any deficiencies in the underlying facts and data go to the weight of Dr. Miller's opinion rather than its admissibility. We disagree. If the dosage of the harmful substance and the duration of exposure to it are the types of information upon which experts reasonably rely when forming opinions on the subject, then the district court was justified in excluding Dr. Miller's opinion that is based upon critically incomplete or grossly inaccurate dosage or duration data. See Soden, 714 F.2d at 506. See also Slaughter v. Southern Talc Co., 919 F.2d 304, 306 (5th Cir.1990) (affidavits of two doctors that "represented nothing more than bare conclusions derived from erroneous data" held inadmissible for summary judgment purposes); Washington v. Armstrong World Indus., Inc., 839 F.2d 1121, 1123-24 (5th Cir.1988) (expert's conclusions "lacked probative value because it was pure speculation based on negative inferences drawn from the testimony of three treating physicians"); Thompson v. Southern Pac. Transp. Co., 809 F.2d 1167 (5th Cir.1987), (expert's opinion on dioxin as source of plaintiff's illness had "insufficient factual basis" because expert had no knowledge of the amount or duration of the exposure).

As we have noted, Rule 703 seeks to ensure that the "facts and data" not otherwise admissible in evidence that form the basis of an expert's opinion are "of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject." Although this rule is primarily directed toward permitting an expert to base his opinion on hearsay or otherwise inadmissible sources, Barrel of Fun, Inc. v. State Farm Fire & Casualty Co., 739 F.2d 1028, 1033 (5th Cir.1984), the inquiry into the "types" of "facts and data" underlying an expert's testimony is not limited to the admissibility of that data. District judges may reject opinions founded on critical facts that are plainly untrustworthy, principally because such an opinion cannot be helpful to the jury.12

The argument that Rule 703 addresses only generic facts and data and is unconcerned with the sufficiency and accuracy of underlying facts as they relate to the case at hand, will lead to the irrational result that Rule 703 requires the court to admit an expert's opinion even if those facts and data upon which the opinion are based are crucially different from the undisputed record. Such an interpretation often will render Rule 703 impotent as a tool for testing the trustworthiness of the facts and data underlying the expert's opinion in a given trial. Certainly nothing in Rule 703 requires a court to admit an opinion based on facts that are indisputably wrong. Even if Rule 703 will not require the exclusion of such an unfounded opinion, general principles of relevance will. In other words, an opinion based totally on incorrect facts will not speak to the case at hand and hence will be irrelevant. In any event such an opinion will not advance the express goal of "assisting the trier of fact" under Rule 702.13

We do not of course say that Rule 703 requires that all facts and data underlying the opinion must relate perfectly to the record facts. As we have pointed out, only when the facts and data are critically inaccurate or incomplete, as determined by what other experts would or would not be willing to base opinions upon, would the facts and data lack the necessary requisites of Rule 703. The district court in this case did not abuse its discretion.

When analyzing the validity of an expert's methodology, we seek to determine whether it connects the facts to the conclusion in a scientifically valid way.14 We answer this question by applying the Frye test: whether the methodology or reasoning that the expert uses to connect the facts to his conclusion is generally accepted within the relevant scientific community.15

In his deposition Dr. Miller stated that the kinds of evidence most often used to establish causation are human epidemiological studies, live animal testing, and in vitro testing. Defendants' experts agreed, but went a step further; they stated that the determination of the pathogenesis of a particular type of cancer requires clearly positive results from one or more of these types of testing. The district court was persuaded that it was inconsistent for Dr. Miller, on one hand, to conclude that these are the main methodologies and, on the other hand, to concede that he did not effectively rely on any of them. That epidemiological, animal, and in vitro studies are, as Dr. Miller said, the primary methods by which medical experts develop their theories of causation is not to say that Miller's methodology was invalid. The Frye question focuses on the proffered methodology alone and looks to the scientific community to determine if general support for that methodology exists.

The critical portion of Dr. Miller's opinion, as it relates to causation, is as follows:

'the same sorts of chemicals and exposures that are associated with small-cell carcinoma of the lung are likely to be associated with small-cell carcinoma elsewhere in the body.'

Christophersen v. Allied-Signal Corp., 902 F.2d 362, 365-66 (5th Cir.1990). Dr. Miller offered no scientific methodology to support this assertion. Dr. Sherwood Gorbach concluded that:

Dr. Miller's presumption that nickel and cadmium have been associated with a certain type of cell in lung cancer and therefore should be associated with a similar type of cell in the colon has no support in medical science and is without foundation.

The other defense expert, Dr. Richard Rudder, held the same view of Dr. Miller's methodology.16 All Dr. Miller had was a scientific hunch, which as far as the record shows, no one else shares. This was enough to support further investigation but was inadequate to support a judgment in favor of Christophersen. The district court found that "Dr. Miller's conclusion that a small cell carcinoma of the lung is likely to be associated with a small cell carcinoma located elsewhere in the body is 'without precedent in cancer epidemiology and is not scientifically correct.' " This finding of what is a scientifically correct conclusion is not for the district court. However, we cannot say that the district court erred in excluding Dr. Miller's testimony. To the contrary, the district court was within its discretion in concluding, albeit implicitly, that Dr. Miller's testimony failed to meet the third threshold test, the Frye test.

Because we find that Dr. Miller's testimony failed to clear either the Rule 703 or the Frye hurdle, we need not consider the district court's application of Rule 403.

The foregoing analysis demonstrates that this was not a "battle of experts" that would be improper for resolution on summary judgment. The district court's ruling that Dr. Miller's opinion was inadmissible was not manifestly erroneous. Because Dr. Miller's testimony was the only evidence of causation, the district court did not err in granting summary judgment for defendants.

AFFIRMED.

The plain words of the carefully created, thoroughly reviewed, fully annotated Federal Rules of Evidence are for courts to follow, not embellish. I am perplexed by the fact that my colleagues in the majority embellish them and in dissent refuse to follow them.

The district court properly excluded expert opinion testimony that it correctly found to be substantially more prejudicial than probative. The majority per curiam approves exclusion, but does so by applying a homemade test for admissibility which not only disregards the plain meaning of the rules, but also builds in a headwind favoring exclusion of such evidence on a far broader basis than the rules permit. Judge Reavley's dissent demonstrates that the majority ignores, in part, and modifies, in part, two of the rules (702 and 703), yet refuses to recognize that the trial court's valid reliance on a third rule (403) requires affirmance. The confusing result of these divergent views could thwart the very reason the court took the case en banc--to light the path district courts should follow in ruling on expert opinion evidence. I hope the result will not be to cause trial judges in this circuit to think we have deprived them of their broad discretion to make fair rulings on the admissibility of such proof. The rules they are bound to follow still control.

Because the district court's ruling under Rule 403 is fully supported by the record, I concur in the result reached by the majority.

I. RULE 702.

Rule 702 defines the basic perimeter of expert opinion evidence:

If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to 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.

Obviously, a medical opinion as to causation would assist the trier of fact in this highly technical scientific case. As Judge Reavley's dissent establishes, it is equally obvious that Dr. Miller is a trained, experienced scientist who possesses specialized knowledge in toxicology and related fields. Both parts of the rule were satisfied. I agree with the district court and Judge Reavley that the summary judgment record establishes that Dr. Miller was qualified to testify as an expert under Rule 702.

The majority errs when it interprets Rule 702 to require the expert's opinion to assist the trier of fact. That is not what the rule says. Expert opinion testimony is admissible when "scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue...." The advisory committee's notes make it clear that this rule focuses any question of assistance on the nature of the jury's factual inquiry rather than on the substance of the expert's testimony.

II. RULE 703.

In two clear English sentences, Rule 703 addresses the permissible evidentiary bases for expert opinion evidence.

The facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to the expert at or before the hearing. If of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject, the facts or data need not be admissible in evidence.

As with Rule 702, the majority pays scant heed to Rule 703's plain language. The result is a confusing and internally inconsistent revision which gives almost no guidance to the district courts except to restrict the admissibility of expert opinion testimony in ways never intended by the Federal Rules of Evidence.

A. Plain Words, Plain Meaning.

The first sentence of Rule 703 allows an expert to base an opinion or inference on facts or data perceived by or made known to the expert at or before the hearing. As the advisory committee's notes explain, this sentence allows three sources for the bases of an expert's opinion: (1) the expert's firsthand observations of facts or data, (2) evidence presented at trial, and (3) presentation of information or data to the expert outside of the courtroom and other than by the expert's own perception. Dr. Miller had no contact with Christophersen. He did not testify at trial. Thus, neither source (1) nor source (2) is applicable. Dr. Miller's testimony falls into the third category.

Rule 703 broadened the ambit of admissible evidence allowed at common law. "The traditional view ... has been that an expert may state an opinion based upon his firsthand knowledge or based upon facts in the record at the time he states his opinion, or based partly on firsthand knowledge and partly on the facts of the record." E. Cleary, McCormick on Evidence Sec. 14, at 31 (2d ed. 1976). This view barred experts from basing their opinions on facts or data presented to them outside of the courtroom other than such as were gained through the expert's own perception. The most significant concern underlying this view was hearsay. McCormick states it this way:

A question is improper if it calls for the witness' opinion on the basis of reports that are not in evidence or are inadmissible as substantive evidence under the hearsay rule (without reciting their contents as hypotheses, to be supported by other evidence as to their truth). The essential objection seems to be that the jury is asked to accept as evidence the witness' inference, based upon someone's hearsay assertion of a fact which is, presumably, not supported by any evidence at the trial and which therefore the jury has no basis for finding to be true.

Id., Sec. 15 at 34.

The second sentence of Rule 703 adopted a new approach to trustworthiness. It allows the expert to rely on inadmissible facts or data if they are of a type reasonably relied upon within the expert's community. See United States v. Williams, 447 F.2d 1285, 1290-91 (5th Cir.1971) (en banc), cert. denied, 405 U.S. 954, 92 S. Ct. 1168, 31 L. Ed. 2d 231 (1972). Trustworthiness of facts or data not tested for admissibility is gained through the assurance that the expert's scientific community reasonably relies on them for the same purpose. See id. at 1290. For example, Rule 703 would permit a doctor to give a diagnostic opinion based upon facts contained in examination or test reports made by hospital technicians even if such reports were inadmissible hearsay, if it is shown that other doctors reasonably rely on such reports when forming similar opinions. Rule 703 says nothing more than that the facts or data need not be admissible in evidence if the reliability inquiry is otherwise satisfied.

If the facts or data are admissible, Rule 703 does not authorize exclusion of the expert opinion. If they are admissible, the inquiry ends, and nothing in Rule 703 authorizes exclusion of the expert's testimony. If they are not admissible, the district court must determine whether the reliability inquiry is satisfied. If it is satisfied, Rule 703 does not authorize exclusion. If it is not, the district court should exclude the testimony. No other reading is consistent with the plain language, history, and purpose of Rule 703.

Both sentences of Rule 703 apply just to the "facts or data" upon which an expert bases an opinion. Rule 703 does not address "methodology"--how the expert uses the facts or data to form an opinion. Rule 703 does not authorize a court to approve or disapprove the expert's conclusion. The words of Rule 703 allow use of facts or data "of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject...." The court's inquiry is not whether experts in the relevant field would reasonably rely on the particular facts or data used by the expert witness. Nor does Rule 703 require a court to determine whether experts in the field would reasonably rely on the same type of facts or data to reach the expert witness's actual opinion. The rule is met if similar experts use facts or data of the same kind to form opinions on the subject in issue. Cf., Soden v. Freightliner Corp., 714 F.2d 498, 503 (5th Cir.1983).

B. A Wrong Turn Followed.

In Viterbo v. Dow Chem. Co., 826 F.2d 420 (5th Cir.1987), this circuit began disregarding the plain language of Rule 703. Viterbo held that expert opinion testimony may be excluded under Rule 703 if, without regard to the admissibility of the underlying facts or data, other experts in the field would reasonably rely on the facts and data assumed by the expert witness. See id. at 422-24. This interpretation is erroneous for two reasons. First, it disregards the fact that the reliability of the facts and data underlying the expert's opinion only comes into question if the facts and data are not admissible. Second, it disregards the fact that Rule 703's reliability inquiry addresses only the "type" of facts and data used by the expert witness and whether experts in the field would reasonably rely on facts or data of that type in forming opinions "upon the subject." This reliability inquiry provides the only necessary and proper guarantee of trustworthiness.

After Viterbo, our cases continued to follow this erroneous construction. See, e.g., Slaughter v. Southern Talc Co., 919 F.2d 304 (5th Cir.1990); Brown v. Parker-Hannifin Corp., 919 F.2d 308 (5th Cir.1990); Washington v. Armstrong World Indus., Inc., 839 F.2d 1121 (5th Cir.1988) (per curiam); Peteet v. Dow Chem. Co., 868 F.2d 1428 (5th Cir.1989), cert. denied, --- U.S. ----, 110 S. Ct. 328, 107 L. Ed. 2d 318 (1989); Orthopedic & Sports Injury Clinic v. Wang Laboratories, Inc., 922 F.2d 220 (5th Cir.1991).

This en banc court errs when it does not overrule Viterbo and its progeny to the extent that they conflict with Rule 703. We should not follow these cases another step.

After the majority's footnote 4 states the correct reading of Rule 703, the opinion abandons the concept that the reliability inquiry only applies when the expert's facts or data are inadmissible. No mention is made of the admissibility of the facts and data upon which Dr. Miller relied. This is a critical mistake because the facts and data he used should have been considered admissible in evidence.

The majority opinion also approves the district court's Rule 703 analysis which criticizes Dr. Miller's facts and data as insufficient support for his conclusion that nickel and cadmium exposure caused Christophersen's cancer. The Rule 703 analysis is not concerned with whether the facts and data relied on by the expert support the expert's actual opinion. Even inadmissible facts or data are tested only to determine whether experts in the field would rely on the type of facts or data relied on by Dr. Miller in forming opinions on the subject of cancer causation.

The majority per curiam responds to this latter criticism with illogic. It reasons that giving plain meaning to the words "type" and "upon the subject" "will lead to the irrational result that Rule 703 requires the court to admit an expert's opinion even if those facts and data upon which the opinion is based are crucially different from the undisputed record." This stops the process too soon. The fact that Rule 703 does not preclude admissibility does not require that an improperly supported opinion must be admitted. Evidence that passes Rule 703 may be excludable under other rules. For example, the opinion might be irrelevant under Rule 401 or substantially more prejudicial than probative under Rule 403.

The majority says that giving the rule its plain meaning "often will render Rule 703 impotent as a tool for testing the trustworthiness of the facts and data underlying the expert's opinion in a given trial." This criticism grafts onto the rule a function that is incompatible with its language and purpose. The trustworthiness aspect of the reliability inquiry has nothing to do with whether the expert's facts or data provide sufficient support for the expert's opinion. Rule 703 does not say that the facts or data upon which an expert witness bases an opinion must supply reasonably reliable support for that opinion. Rather, the rule treats the reliability inquiry as a sufficient guarantee that an expert's inadmissible facts or data are sufficiently trustworthy to overcome the reasons why they are inadmissible. The rules deal with fundamentally unsupported but relevant expert opinions only in terms of probity versus prejudice under Rule 403, discussed below.

C. Rule 703 Applied.

In today's case, the facts and data upon which Dr. Miller based his opinion were the Manoliu affidavit, Christophersen's medical records, and medical literature. Judge Reavley's dissent sets them out in careful, complete, and correct detail. In rendering summary judgment, the district court did not expressly determine that these facts and data could not have been admitted in evidence. For summary judgment purposes, the facts contained in Manoliu's affidavit should have been considered admissible at trial in the form of Manoliu's direct testimony. The facts contained in the medical records should have been considered admissible in the form of direct testimony by those who made the records or as records of regularly conducted activity. See FED.R.EVID. 803(6). Medical records also might be the type of sources of information upon which cancer experts reasonably rely when forming their opinions as to the causes of a person's cancer. No competent summary judgment proof suggests that they were not. The data contained in the medical literature would have been admissible over a hearsay objection under the learned treatise exception to the hearsay rule. See FED.R.EVID. 803(18). The majority and the district court did not consider whether the facts and data relied on by Dr. Miller were admissible.

The district court also attempted to determine whether experts in the field of cancer research would have reasonably relied on the particular facts and data used by Dr. Miller (the Manoliu affidavit, medical literature, and medical records) to form his actual opinion (that nickel and cadmium caused Christophersen's colon cancer). This analysis was also improper. Rule 703 only asks the court to determine whether experts in the field of cancer research would have reasonably relied on facts or data of this "type" in forming opinions or inferences "upon the subject" of cancer causation.

Insistence on a punctilious observance of the intended operation of Rules 702 and 703 does not put form over substance. Rather it enforces the spirit of the Federal Rules of Evidence which provide the trier of fact with all but a narrow band of relevant evidence. Evidence authorized by the literal terms of Rules 702 and 703 may not be excluded unless the court balances probity against substantially greater prejudice as required by Rule 403, discussed below. The majority opinion destroys the value of that weighing because it allows district courts to exclude expert opinions under Rules 702 and 703 for reasons those rules do not permit.

Because the district court's analysis of Dr. Miller's testimony was premised on an incorrect legal interpretation of Rule 703 that would deprive a jury of evidence it should be able to consider, I agree with Judge Reavley that exclusion on the basis of Rule 703 was manifestly erroneous and that the majority erred in basing affirmance on this rule.

III. FRYE.

I completely agree with Judge Reavley's analysis of the limited history of Frye in this circuit and criticism of its adoption here. I support his views with three additional points. First, if Frye is a rule of evidence, it has not survived the enactment of the Federal Rules of Evidence. See 3 J. Weinstein & M. Berger, Weinstein's Evidence p 702, at 702-36 (1990); 22 C. Wright & K. Graham, Federal Practice and Procedure Sec. 5169, at 108 (1978). Second, if it is a substantive rule, its adoption in diversity cases is foreclosed by Erie. Until today, this circuit has limited Frye to a narrow class of federal criminal cases to which Erie does not apply. Third, Frye is neither a good rule nor one the Court must adopt to decide this case. Rather than extending Judge Reavley's critique of Frye, I simply note that the three leading treatises on the law of evidence support my position that Rule 403 is the better test. See E. Cleary, McCormick on Evidence Sec. 204, at 491 (2d ed. 1976); 3 J. Weinstein & M. Berger, Weinstein's Evidence p 702, at 702-34-44 (1990); 22 C. Wright & K. Graham, Federal Practice and Procedure Sec. 5168, at 86-91 (1978).

IV. RULE 403.

In clear declaratory language, Rule 403 grants a limited power to exclude evidence.

Although relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, confusion of the issues, or misleading the jury, or by considerations of undue delay, waste of time, or needless presentation of cumulative evidence.

With a footnote, the majority dismisses the pertinence of Rule 403 to today's case. The district court's application of Rule 403 is essential to the outcome of the appeal. Individual rules cannot be read in isolation. Rule 403 is as much a part of the Rules of Evidence that govern this case as Rules 702 and 703. If it were not, I would agree with Judge Reavley and join his dissent. However, since it applies as strongly to expert opinion testimony as it does to any other evidence, see 1 J. Weinstein & M. Berger, Weinstein's Evidence p 403, at 403-5 (1989), I would affirm the district court's decision to rely on it to exclude Dr. Miller's testimony.

Rule 403 allows a court to exclude relevant evidence when its probative value is substantially outweighed by the danger of unfair prejudice. An expert's opinion may be based on such erroneous facts or data, such proven unsound methodology, or such internally inconsistent reasoning that its probative value is minimal. I agree with the majority's footnote 10 which acknowledges that expert testimony may create a special kind of prejudice. When an opinion, especially one a lay person finds as arcane and speculative as cancer causation, is based on erroneous data, reasoning, or methodology, qualifying the opining witness as a medical expert carries a likely danger than the opinion will be substantially more prejudicial than probative. The fact that a witness is labeled an "expert" under Rule 702 would certainly be an improper basis for a jury's decision to believe the witness' opinion.

These principles introduce no new concepts into our jurisprudence. While I reject Viterbo 's departure from the plain meaning of Rule 703, I agree with its reasoning that if an opinion is fundamentally unsupported, then it offers no expert assistance to the jury; and that lack of reliable support can render an opinion substantially more prejudicial than probative, making it inadmissible under Rule 403. Viterbo, 826 F.2d at 422; see also Washington v. Armstrong World Indus., Inc., 839 F.2d 1121, 1123-24 (5th Cir.1988) (per curiam).

Judge Reavley's dissent maintains that, by excluding Dr. Miller's testimony, the district court took a fact issue from the jury. An analysis of probity versus unfair prejudice almost always depends to some extent on facts, but where undisputed proof supports the district court's conclusion as it does here, the issue is one properly decided by the court. When a permissible evidentiary ruling, supported by undisputed facts, excludes evidence, it takes away nothing that is proper grist for a jury.

In today's case, the undisputed record confirms that Dr. Miller relied on erroneous factual assumptions. Dr. Miller formed his opinion and testified at his deposition without any accurate information on Christophersen's exposure. Dr. Miller admitted that a person's level of exposure to possible carcinogens is critical. The studies on which he relied indicate that carcinogens exhibit a dose-response effect. Dr. Miller's only source of information on exposure was Manoliu's affidavit. Significantly, Manoliu had no idea how much, if any, nickel and cadmium was present in the fumes. Dr. Miller formed his opinion and testified at his deposition without any knowledge of what particles or compounds were present in the fumes or in what concentration, the size of the impregnation and soak rooms where the fumes originated, and the physical structure and ventilation in the plant and Christophersen's office. Dr. Miller's only basis for his assumption that the fumes contained any nickel and cadmium at all was Manoliu's bare assertion that these compounds were present in the fumes because the fumes were irritating. When Dr. Miller was confronted with the fact that he had formed his opinion without any credible exposure data, his sole response was that he would not change his opinion unless it was shown that the fumes contained absolutely no nickel and cadmium. He failed even to attempt to justify this unique view which was at odds with his own view that the level of exposure was critical and contrary to the dose-response effect detailed in the studies on which he relied.

Dr. Miller's reasoning and methodology were also seriously deficient. He asserted that nickel and cadmium caused Christophersen's small cell colon cancer because they have been associated with small cell cancer of the lung despite the fact that he could cite no authoritative sources for this type of associative reasoning. The uncontradicted summary judgment evidence indicated that, without additional support, Dr. Miller's presumption that nickel and cadmium should be associated with cancerous cells in the colon because they have been associated with a similar type of cancerous cell in the lung is without precedent in cancer epidemiology and has no foundation in medical science. Moreover, Dr. Miller agreed that the determination of the pathogenesis of a particular form of human cancer requires human epidemiological studies, animal studies, and in vitro testing. He conceded that he had never seen an epidemiological or animal study demonstrating a causal association between exposure to nickel and/or cadmium and colon cancer.1

I do not disagree with the dissent's view that Dr. Miller's testimony passed Rules 702 and 703. I very much disagree with its view that the court was required to admit his opinion testimony because the provisions of these two rules were met. The trial court was requested to review the evidence under Rule 403. He did so and concluded that Dr. Miller's testimony should be excluded.

Under the undisputed portions of the summary judgment record, the trial judge could find that Dr. Miller's opinion was supported only by his credentials and his persistent refusal to acknowledge the inadequacy of his methodology. The district court determined that the probative value of Dr. Miller's testimony was substantially outweighed by the danger of unfair prejudice. To reverse this Rule 403 ruling, we must find manifest error. There was none.

V. CONCLUSION.

The majority opinion distorts or ignores the letter and spirit of the Federal Rules of Evidence. Today's case ought to be decided based on the plain language of the three applicable rules. By overlaying these rules with its own agenda for exclusion, the majority disserves our trial judges in this difficult area of their work. The dissent's refusal to acknowledge that the district court should be trusted to weigh probity against substantial unfair prejudice is equally in error. Because I cannot join either opinion, I respectfully concur only in the affirmance of the appealed judgment.

REAVLEY, Circuit Judge, with whom KING, JOHNSON and WIENER, Circuit Judges, join, dissenting:

The judges of this court have in recent years been sending warning signals about their displeasure with expert testimony.1 Today the court "takes hold" of expert testimony by taking over. The per curiam opinion effectively allows judges to decide the reliability, weight, and relative merit of expert opinions, at least in toxic tort cases. And with such control, we signal a willingness to increase the proof and persuasion burdens of the disfavored party. The author of the per curiam claims to "introduce no new concepts to our jurisprudence." Surely my colleagues know better, or at least they should know that their use of these concepts confuses the admissibility of evidence with the sufficiency of evidence, changes the rules of evidence without benefit of amendment, denies Mrs. Christophersen her right to trial by jury, and eliminates substantive rights in tort cases where federal courts have only diversity jurisdiction.

The en banc court perpetuates the district court's lack of study or appreciation of the record. Lengthy depositions were taken of three witnesses for the plaintiff. Dr. Waymon Johnston, an industrial engineering professor, testified to the negligence of the defendants in failing to warn of the deadly hazard of nickel and cadmium exposure. Christophersen's co-worker, Edgar Manoliu, testified about the conditions at the plant where Christophersen worked and his exposure to nickel and cadmium. And Dr. Miller testified that in his opinion it was that exposure which caused the colon cancer leading to Christophersen's death. The defendant presented four affidavits of medical experts who disagreed with the opinion of Dr. Miller and denied that scientific proof exists that cadmium and nickel fumes can cause small-cell cancer in the colon. These affiants for the defendants did not say that science has disproved the opinion of Dr. Miller; they did not even address his discussion of links between toxic carcinogens and genetic abnormalities; and they did not submit to depositions and the revealing cross-examination to which seven defense attorneys subjected Dr. Miller during his two-day deposition. Upon that basis and without recourse to a jury, the district court and this court choose to accept the position of the defendants' doctors and deny Mrs. Christophersen a trial.

A. The Affidavit and Deposition of Edgar Manoliu

Edgar Manoliu worked at Marathon's manufacturing plant in Waco from 1972 until he retired from his position as production supervisor in 1985. He stated that Albert Christophersen, a design and applications engineer, officed in the main administration building from 1972 to 1980, and that his duties required him to visit the main production area two or three times per week for up to half an hour per visit. In 1980, the company assigned Christophersen to an office in the main production area of the plant, located in an enclosed cinder-block loft area, approximately sixty yards from the impregnation and soak room. Christophersen's windows and door generally remained open. According to Manoliu the manufacturing plant and the impregnation and soak room contained "many fumes and gases" which "included heated fume and airborne particles of cadmium and nickel alloys." Christophersen's visits to the impregnation and soak room--the site of the heaviest concentration of nickel and cadmium fumes--continued at the rate of two to three times per week.

Manoliu did not testify to the precise composition or level of the fumes and dust. His expertise did not extend so far. Basing his testimony on personal knowledge and observation, Manoliu employed neither mathematics nor chemistry, but nevertheless adduced significant evidence of exposure. Christophersen's exposure occurred chiefly from visits to the impregnation and soak room, and from dust and fumes pervading the plant and his open office. Manoliu testified to a heavy concentration of nickel and cadmium fumes in the impregnation and soak room, the doors of which opened frequently. The impregnation and soak room process involved repeatedly dipping plaques on racks into bins of cadmium and nickel solutions for one to two minutes, then raising the racks to allow dripping. After dripping, the racks entered the dryers. Manoliu testified that the fumes emerged from the bins of cadmium and nickel solutions, as well as the dryers, and that lifting the racks out of solution created a "tremendous" release of fumes. He testified that all employees in the impregnation and soak room complained of the intolerable atmosphere, and attributed their scratchy throats and noses to the fumes (which were visible clouds) emerging from the soak baths and the dryers. Visitors to the area also experienced physical reactions.

Manoliu identified several sources of the fumes and dust that pervaded the plant and reached Christophersen's office through his open door and windows. Besides the concentrated fumes that emanated from the impregnation and soak area, the shearing of nickel and cadmium plaques created cadmium and nickel dust by cutting through "active areas" of nickel and cadmium plaques. The Waco plant, unlike other Marathon plants, did not have a vented exhaust system in the shearing area. Manoliu testified that workers in the shearing area eventually had to wear masks, and that Christophersen sometimes walked through the area.

The plaque-making area also generated significant quantities of nickel dust. Employees there had to wear small dust masks, but Manoliu testified that when they removed the masks, he observed dust "all over their noses," and that he reported this inadequacy to a plant authority, who agreed to "look into it." No change occurred. Additionally, the tab room, which involved removal of residues with steel brushes, generated significant quantities of cadmium dust. This area was not enclosed, and Manoliu testified that "there was no exhaust system whatsoever."

Periodic problems with the scrubbers or the roof fans, which failed to pull excess fumes out of the plant, compounded the problem. Damp, muggy days created the worst conditions. Even if the ventilation system worked properly, high humidity prevented the exhaust system from carrying off sufficient quantities of the noxious dust and fumes.

Manoliu further testified that nickel and/or cadmium spills--involving tank overflows of five to fifteen gallons onto the floors--occurred about once a month near the impregnation tanks. Pipe leaks, during transfer of nickel nitrate and cadmium nitrate from trucks to the soak room, likewise occurred about once a month.

To the foregoing description of toxic exposure, Manoliu added the cancer toll affecting several Marathon friends and co-workers in the previous five to ten years. In his affidavit, he named eight people diagnosed or dead with cancer, including Albert Christophersen. In his deposition, Manoliu noted that his list of cancer victims included only people he personally knew, that he never investigated the matter, that he named only supervision and white-collar workers such as himself and Christophersen, and that he knew nothing about the medical histories of hourly employees. Manoliu is himself a cancer victim.

B. The Affidavit and Deposition of Dr. Lawrence Miller

Dr. Miller stated that he reviewed Christophersen's medical records and autopsy report, as well as numerous medical articles and research materials addressing the effect of nickel and cadmium on humans.2 He also relied on his extensive educational and clinical background, and on information about Christophersen's work history and toxic exposure supplied by Manoliu. Dr. Miller concluded, based on reasonable medical probability, that the metastatic cancer which killed Christophersen was caused by his exposure to toxic nickel and cadmium fumes.

Despite "the Rule 702 concerns expressed by the district court," Dr. Miller is a board-certified specialist in internal medicine, pulmonary disease, and critical care medicine. His pulmonary disease specialty, primarily lung disorders, includes two years of patient care. He also notes specialization in epidemiology and toxicology, and he holds a Masters of Public Health in toxicology. His background in oncology includes courses at Harvard Medical School and internal medicine training, which involved substantial oncology exposure. He is not an oncology specialist, but oncology constitutes a subspecialty of internal medicine, in which he does specialize. Dr. Miller notes that an oncologist is more qualified in the treatment of cancer, but not necessarily more qualified concerning carcinogenesis. By the time of his deposition, Dr. Miller knew that his contract with Tufts University provided that all fees received for consultation services would go to the university.

Albert Christophersen had a villous adenoma which underwent a malignant transformation to a small-cell type of histology. His small-cell colon cancer then metastasized to his liver. Dr. Miller concluded, as follows, that toxic exposure to nickel and cadmium produced Christophersen's cancer.

a. Nickel, Cadmium, and Small-Cell Cancer

Nickel and cadmium cause cancer. The types of evidence honored by all of the experts--epidemiology, animal studies, and in vitro studies--establish the carcinogenicity of nickel and cadmium.3 But did nickel and cadmium cause Christophersen's colon cancer?

Science so far has established specific links of nickel and cadmium to lung, prostate, and renal cancers.4 Villous adenoma small-cell cancers, which Christophersen had, are quite rare, and are associated with several toxic exposures. Research discloses an association between nickel and cadmium and small-cell lung cancer.5 Small-cell carcinoma appears to be the same cell (identical histology) regardless of its location in the body, which is why medical literature does not necessarily address specific organ systems in discussing small-cell cancer. Dr. Jacqueline L. Torell, who performed the autopsy the day after Christophersen's death, described the tumor in his colon as being "morphologically quite similar to small-cell undifferentiated carcinoma most frequently arising in the lung." She noted that a few of the clusters had a somewhat papillary arrangement but there was not true gland formation or mucin typical of colon cancer. Recent medical evidence relates small-cell cancer to abnormal alterations of genetic materials.6 One specific mechanism of nickel and cadmium carcinogenicity is alteration in DNA synthesis. From these findings, Dr. Miller concludes that nickel and cadmium can cause small-cell colon cancer.

b. Christophersen's Exposure

The district court and this en banc court have relied heavily upon the notion that the record lacks proof of exposure of Christophersen to toxic fumes. Of course, neither Manoliu nor Dr. Miller know the precise level of that exposure, and the lack of quantified dose and exposure levels could handicap Mrs. Christophersen's case. But this imprecision does not justify exclusion of Dr. Miller's opinion. Seven defense attorneys cross-examined Dr. Miller at length about disputed facts upon which his opinion was based. The questions about exposure raised by our courts were put to Dr. Miller, and his opinion of causation remained unchanged. He readily agreed that dose and exposure levels are important, but his specific conclusion in Christophersen's case rests on sufficient exposure to nickel and cadmium, not precisely known exposure. He testified that his opinion remained valid "as long as the exposure occurred in general as described to me."7 None of the defendants, including the employer, presented any proof concerning the fumes and dust in the plant, or Christophersen's exposure.

c. Alternative Causation

Dr. Miller considered the various possibilities of alternative causation. Christophersen was not a smoker. Mrs. Christophersen smoked, but Dr. Miller considered this source insufficient for Christophersen's small-cell colon cancer. Nor was Christophersen a heavy drinker. Dr. Miller noted that associations of alcohol and gastrointestinal carcinoma generally occur with chronic alcoholics. Dr. Miller found no evidence of exposure to asbestos, nor was there evidence of hereditary or family associations of small-cell cancer. While evidence exists that diet contributes to the development of adenocarcinomas of the colon, no such evidence links diet to small-cell cancers.8 Dr. Miller stated that "[d]ifferent types of cancers have different biochemical, cellular characteristics, natural histories and etiologies, and one can't just necessarily lump them like that."

We assign to the trial court the threshold determination of the expert's qualifications, whether the expert's testimony will assist the jury, and whether the facts assumed by the witness as the basis for the opinion are supported in the evidence so as to make that opinion relevant. None of these inquiries warrants exclusion of Dr. Miller's opinion.

A. The Neglected Virtues of the Adversarial System

In a case where the plaintiff claimed that his disease was caused by his toxic work environment, and presented questionable expert medical testimony in support thereof, this circuit declared en banc that "it is the function of the jury as the traditional finder of the facts, and not the Court, to weigh conflicting evidence and inferences, and determine the credibility of witnesses." Boeing Co. v. Shipman, 411 F.2d 365, 375, 377 (5th Cir.1969) (en banc) (affirming denial of directed verdict and judgment n.o.v. even "[t]hough the expert medical proof is not entirely reliable in determining the causal connection between Shipman's working conditions and his ailments"). I need not buttress with a string citation this once well-settled proposition. We properly entrust determinations of evidentiary weight and credibility to the jury--even in "scientific" cases--because of our faith in the adversarial process.

If the evidence is relevant, and not clearly wrong, it goes to jurors, who then "have the benefit of cross-examination and contrary evidence by the opposing party." Barefoot v. Estelle, 463 U.S. 880, 898, 103 S. Ct. 3383, 3397, 77 L. Ed. 2d 1090 (1983) ("[T]he rules of evidence generally extant at the federal and state levels anticipate that relevant, unprivileged evidence should be admitted and its weight left to the factfinder, who would have the benefit of cross-examination and contrary evidence by the opposing party.").9 The Supreme Court resolutely championed the adversarial process upon considering the admissibility of psychiatric testimony in predicting future dangerousness. Despite expert testimony--purportedly representing the "best" clinical research to date--that the accuracy of such prediction reached no more than one in three, the Court declared:

All of these professional doubts about the usefulness of psychiatric predictions can be called to the attention of the jury. Petitioner's entire argument ... is founded on the premise that a jury will not be able to separate the wheat from the chaff. We do not share in this low evaluation of the adversary process.10

The Federal Rules of Evidence "embody a strong and undeniable preference for admitting any evidence having some potential for assisting the trier of fact and for dealing with the risk of error through the adversary process." DeLuca v. Merrell Dow Pharmaceuticals, Inc., 911 F.2d 941, 956 (3d Cir.1990); accord In re Paoli, 916 F.2d at 857 (noting that "in making reliability determinations, courts must err on the side of admission rather than exclusion"). The majority's usurpation of critical evidentiary weight determinations plainly disregards the virtues of the adversarial system and the precedents supporting it.

Courts may, of course, exclude evidence that is patently unreliable and therefore offers no assistance to the jury. The language expressing such patent unreliability varies, but never signals testimony or assumptions that are merely controversial, debatable, questionable, unsettled, or suspicious. These terms connote weight and credibility. Instead, courts speak of testimony that is "almost entirely unreliable,"11 reliance upon "assumptions devoid of any basis in the real world,"12 opinions that are "abusive of the known facts"13 or "contrary to the proven facts,"14 or so manifestly wrong as to offend common sense.15 If the record establishes a critical fact contrary to the expert's testimony, or if a court may take judicial notice of a fact that fatally contradicts the assumptions of an expert, then his or her testimony ought to be excluded.16

No such infirmity warrants exclusion of Dr. Miller's testimony. Nor do defendants present any wealth of contrary evidence, of the sort that produced the Agent Orange17 decision often cited for a restrictive view of admissibility. Compared to substantial research and experimentation conducted on Agent Orange, the accumulated knowledge--through traditional methods--of the relationship between nickel, cadmium, and small-cell colon cancer is scant.18 The causation issue is plainly not sufficiently investigated to warrant summary proceedings based on direct judicial precedent or judicial acknowledgement of decisive negative findings. No clearly established facts in the record controvert the scientific facts informing Dr. Miller's reasoning: the carcinogenicity of nickel and cadmium and their capacity to reach the colon; the unique characteristics of small-cell carcinoma and its relative rarity in the colon;19 the linkage of small-cell carcinoma to toxins such as nickel and cadmium; and the genetic-alteration mechanism associated with both small-cell carcinoma and nickel and cadmium.

To dispatch the controversial question of causation, the district court simplistically characterized Dr. Miller's testimony as grounded exclusively in similar cell appearance, and simply chose to believe the opposing experts who issued broad and incomplete rebuffs of Dr. Miller's analysis. Cf. In re Paoli, 916 F.2d at 858 (district court made improper credibility judgment in choosing to "believe[ ] defendants' experts that ... meta-analysis was not reliable, and disbelieve[ ] plaintiffs' experts who said that it was"). For this arrogation of weight and credibility determinations, the district court and the majority enjoy the force of neither precedent nor principle. Whether Dr. Miller's testimony would prevail as proof of causation belongs to the jury.

To the foregoing, it may be countered that a "battle of experts" on a point of hard science demands the impossible from a lay jury. If the experts cannot agree, the argument goes, how can we expect a jury to resolve the question?20 Yet by our charge to the jury,21 we routinely entrust difficult determinations to jurors that assume their possession and use of critical capacities.22 Indeed, the argument simply acknowledges the mandate of our legal system: to resolve conflicts, even seemingly intractable conflicts, from which we cannot retreat when parties invoke the machinery of the legal system. In this sense, the jury represents neither the ideal arbitration of scientific conflict, nor its permanent resolution, but simply the essential voice of the community in solving one problem fairly brought before it.23 Let the experts settle the larger dispute in due time; we have cases to resolve.

In this case, we do not demand of a jury any exotic sophistication. The medical explanations may employ unfamiliar terms, but the resolution of the evidentiary conflict demands only attentive common sense. With its deference to summary rendering of Christophersen, the majority endows judges with the work of juries.

B. Current Ways of Taking Hold Without Taking Over

Particular trials beget particular needs. Judges with concerns about scientific evidence have at their disposal many mechanisms that do not distort the Federal Rules of Evidence, such as appointing (or allowing the parties to negotiate for) neutral experts to assist in comprehending complex issues,24 referring issues to special masters, employing technical advisors, presiding over pretrial colloquia with experts to clarify issues and narrow disagreements, requiring experts to follow certain protocols,25 providing jurors with glossaries, allowing civil jurors to question experts, issuing special instructions to eliminate some of the mystery of technical issues,26 or bifurcating the trial to resolve the issue of causation without exploring conduct or damages.27

If current mechanisms prove inadequate to the general problem perceived by the majority, it remains the duty of courts to follow either precedent or the proper procedures of reform.28 The American Law Institute, for example, is presently considering a proposal for "Blue Ribbon Science Panels" and a "Federal Science Board" to assist courts in the disposition of scientific questions.29

The majority regimen short-circuits present mechanisms and proposals for modest reform by effectively converting judicial chambers into science star chambers. We need practical approaches to the intersection of law and science, not disguised control that skews the system toward exclusion of otherwise relevant evidence.

The district court analyzed Dr. Miller's "methodology" under Rule 703, which neither the Federal Rules nor the majority opinion sanctions. The language of Rule 703 governs an expert's use of "facts or data ... of a type reasonably relied upon by experts," not an expert's opinion or "methodology."30 This misapprehension of Rule 703's scope does not justify the majority's magnanimous deference to the district court's ruling.31 Yet the majority merely refurbishes that erroneous framework, shifting the district court's Rule 703 analysis to the Frye rubric, and then, alluding to its limited scope of review, generously infers that the district court's inaccurate and incomplete portrayal of Dr. Miller's testimony was "sufficiently broad to be understood as rejecting Dr. Miller's methodology."

The majority opinion quietly corrects the district court's misuse of Rule 703, but in holding that the district court did not abuse its discretion in rejecting Dr. Miller's reliance on "critically inaccurate or incomplete" data, the majority commits error of its own. The majority emphasizes that Dr. Miller's information on Christophersen's exposure to nickel and cadmium came from the Manoliu affidavit, which did not precisely quantify dose and exposure levels.32 Further, my colleagues seem impressed by the allusion in Dr. Miller's affidavit to twenty years of exposure, while Manoliu's affidavit establishes that Christophersen's exposure at the Waco plant lasted a mere fourteen years.33 These defects make marvelous trial arguments, but badly mistaken bases for exclusion under Rule 703.

An expert may base an opinion on "facts or data" otherwise inadmissible, if "of a type reasonably relied upon by experts in the particular field in forming opinions or inferences on the subject."34 Courts properly defer "to the expert's view that experts in his field reasonably rely on such sources of information." Greenwood Utilities Comm'n v. Mississippi Power Co., 751 F.2d 1484, 1495 (5th Cir.1985); see also In re Japanese Electronic Products, 723 F.2d at 277 (district court misinterpreted Rule 703 by "substituting its own opinion as to what constitutes reasonable reliance"). The district court's erroneous invalidation of Dr. Miller's opinion--based upon deficiencies that Dr. Miller expressly considered and dispatched in his deposition--exceeds the plain language of Rule 703 and the deference due an expert's reliance on chosen facts and data. The ten cases string-cited by the majority do not suggest a different interpretation. These patently distinguishable cases all address questions of evidentiary sufficiency rather than admissibility,35 or circumstances of plain incompetence to testify to the pertinent issue,36 or reliance on critical facts contrary to established facts.37

Three possibilities occur here, and none eliminates Dr. Miller's opinion at the summary judgment stage. First, if the record shows that Dr. Miller based his opinion on critical assumed facts that do not exist in Christophersen's case, then the opinion is irrelevant. The proffered evidence runs afoul of Rule 401 as well as Rule 702, rather than Rule 703. But this record does not discredit the factual basis for Dr. Miller's opinion. Second, if we have a factual dispute as to the nature of Christophersen's exposure and whether it suffices for what Dr. Miller requires to justify his opinion, the opinion is conditionally relevant. But that dispute goes to the jury under Rule 104(b), not to the judge. Third, if Dr. Miller and the defense experts disagree on the extent of exposure required to cause cancer, we simply have a question of fact to be weighed and decided by the trier of facts.

If the district court or the majority feels that Dr. Miller's reliance on imprecise dose and exposure levels rendered his opinion tenuous, then our cases dictate the procedure to follow: "It is the jury's province to determine whether expert testimony rests upon tenuous evidentiary inferences and weak scientific or engineering data." Mitchell v. Lone Star Ammunition, Inc., 913 F.2d 242, 252 (5th Cir.1990). No record evidence establishes that precise dose and exposure quantification must precede any causation opinion, only that such precision helps. If there be weakness in his reliance on imprecise data, the determination belongs to the jury.38 The district court's ruling to the contrary was manifestly erroneous.

This circuit has continued to apply the rule of Frye v. United States39 to determine admissibility of polygraph examinations, post-hypnotic testimony, and voice stress analysis.40 Now the Fifth Circuit, without precedent, reaches beyond novel device or technique and subjects expert reasoning to judge-determined reliability.

While the Fifth Circuit has never joined the chorus of Frye detractors,41 the six Fifth Circuit cases that actually cite Frye signify the limited purpose of that case.42 We have only once employed Frye outside the criminal context,43 never applied it to "reasoning," and indeed once expressly limited the Frye doctrine to "pseudo-scientific data."44

Until today, we soundly limited the Frye doctrine to particular techniques, "novel scientific evidence,"45 that reflect the factual context of Frye. To my knowledge, the Fifth Circuit has never excluded otherwise relevant evidence strictly on the basis of lack of general scientific acceptance. To do so would impose a significantly heavier burden of proof and persuasion on the offering party. See Peteet v. Dow Chem. Co., 868 F.2d 1428, 1433 (5th Cir.1989) ("[T]he absence of a scientific consensus on a given theory does not affect the admissibility of an expert's opinion."), cert. denied, --- U.S. ----, 110 S. Ct. 328, 107 L. Ed. 2d 318 (1990); Osburn, 825 F.2d at 915 (expert's opinion need not be generally accepted in scientific community to be sufficiently reliable and probative).

The district court of course did not cite to Frye or employ any kind of Frye analysis. The doctrine is doubtless alien to it in such a context. But the majority found that "the district court was within its discretion in concluding, albeit implicitly, that Dr. Miller's testimony failed to meet the third threshold test, the Frye test." How did they get there? I do not know, but they speak of "methodology."

If his study and the reasons for his opinion are Dr. Miller's "methodology," nothing distinguishes his methods from those of other experts. He examined the available information and literature, applied his experience and education, and gave an opinion with his reasons. If his reasons are his methodology, then here it is: (1) Christophersen had small-cell colon cancer that metastasized to his liver; (2) nickel and cadmium cause cancer; (3) some evidence suggests that nickel and cadmium act synergistically to produce cancer; (4) medical science so far has established specific links to lung, prostate, and renal cancers; (5) small-cell carcinoma appears to be the same cell (identical histology) regardless of its location in the body; (6) one of the physicians who examined Christophersen's colon biopsy material believed it was virtually identical to tumors of lung origin; (7) substances designated as "tumor promoters" are associated with a wide range of cancers, and there is some evidence that nickel is a tumor promoter; (8) one mechanism of nickel and cadmium carcinogenicity is alteration in DNA synthesis; (9) medical evidence relates small-cell cancer to abnormal alterations of genetic material; and (10) small-cell carcinomas, which are quite rare, are associated with toxic exposures.

None of the foregoing propositions abuses facts known to the district court or the majority. None is contrary to established scientific knowledge. Nor do we know specific cancers invariably associated with nickel and cadmium, or specific causes invariably associated with small-cell cancer.46 The only evidence in the record to substantiate the majority's Frye holding is the contrary conclusion of the defense experts, who do not even address critical aspects of Dr. Miller's disputed "methodology." And even had they refuted his reasoning with withering particularity, upon what basis can this majority infer the lack of "general acceptance," especially when the district court did not even address that issue? Are the affidavits of experts hired by the defense conclusive proof of "general acceptance"? Thus does the majority manage to introduce a new regimen, and simultaneously provide a case-study of its mischievous abuse.

"[T]his case involves a situation where science has some meaningful information but scientific 'truth' has not so completely hardened as to prevent legitimate difference of true expert opinion in a particular concrete context." Osburn, 825 F.2d at 915 & n. 10 (rejecting argument that testimony of leukemia victim's experts should have been excluded because it conflicted with widely-held theory that chloramphenicol cannot cause leukemia without first causing aplastic anemia, which plaintiff never had). Our tradition, our cases, and our rules properly deliver this case to the jury. It is pretense to say that rules and precedent support the per curiam opinion, to claim that it is consistent with the rights of Mrs. Christophersen under the Seventh Amendment and Texas law, or to say that it is anything other than judicial fact-finding.

Chief Judge Clark identifies the per curiam errors on Rules 702 and 703 but then does greater mischief with Rule 403. He misreads this record and misuses Rule 403.

Judge Clark writes that the "undisputed record" disproves Dr. Miller's factual assumptions. If that were so, Dr. Miller's opinion about facts unrelated to this case would not be relevant. And the opinion would not assist the fact finder. As I have said above, the opinion would be excluded under settled law. But how does this record disprove Dr. Miller's assumptions about Christophersen's exposure to toxic fumes? And how can Judge Clark enshrine as established fact that Dr. Miller delivered his opinion "without any accurate information on Christophersen's exposure"? One fact is truly uncontradicted: Christophersen was regularly exposed to the fumes and dust of this nickel-cadmium battery plant for over 13 years. Dr. Miller said he had enough information to support his causation opinion. That is the state of the record. By what rule or reason do we leap to judgment?

True, defense attorneys confronted Dr. Miller with all varieties of skepticism about Christophersen's exposure. But Judge Clark discloses his chosen result by converting defense attorney skepticism into the "fact that [Dr. Miller] formed his opinion without any credible exposure data." Then Judge Clark caricatures Dr. Miller as a man oblivious to conflicting evidence. Dr. Miller maintained that Christophersen's exposure was sufficient to cause, and in his opinion did cause, Christophersen's cancer and death. He did not say his opinion would be different only if there were no nickel and cadmium fumes; he simply stated that he might change his opinion if Christophersen had visited the soak room only once a month, and that his conclusion would certainly be different if there had been no exposure to toxic fumes.

The causal connection between these toxic agents and small-cell colon cancer is disputed. The ordinary routes of epidemiology, animal studies, and in vitro studies establish the carcinogenicity of nickel and cadmium, but do not yet resolve with scientific certainty--one way or the other--the link between these carcinogens and the rare small-cell colon cancer. My colleagues presume the lack of connection and predicate summary judgment on that judicial presumption. They disregard the different standards of law and science. Courts do not require scientific certainty. An opinion about probability suffices. The fact issue stands.

If Judge Clark would override that fact issue with Rule 403, he fashions a rule with the following new and drastic effect. Judges may weigh contradictory evidence and exclude any proffered evidence considered unreliable. The judge's opinion about the contested evidence determines unreliability. The force of the proffered evidence on the very point at issue becomes the prejudice.

The advisory committee notes to Rule 403 define "unfair prejudice" as "an undue tendency to suggest decision on an improper basis." What is the improper basis or prejudice here, unless a judge simply assumes that Dr. Miller is wrong and the defendants' affiants are right? That assumption has not been a predicate in the past for Rule 403 exclusion. See Dollar v. Long Mfg., N.C. Inc., 561 F.2d 613, 618 (5th Cir.1977), cert. denied, 435 U.S. 996, 98 S. Ct. 1648, 56 L. Ed. 2d 85 (1978). We have viewed Rule 403 as an extraordinary remedy to be used cautiously and sparingly. United States v. McRae, 593 F.2d 700, 707 (5th Cir.), cert. denied, 444 U.S. 862, 100 S. Ct. 128, 62 L. Ed. 2d 83 (1979); United States v. Thevis, 665 F.2d 616, 633 (5th Cir. Unit B), cert. denied, 459 U.S. 825, 103 S. Ct. 57, 74 L. Ed. 2d 61 (1982). Others have agreed. United States v. Cross, 928 F.2d 1030, 1048 (11th Cir.1991) ("The 'major function' of Rule 403 is 'limited to excluding matter of scant or cumulative probative force, dragged in by the heels for the sake of its prejudicial effect.' ") (quoting Thevis and McRae ); DeLuca, 911 F.2d at 957 (if "testimony survives the rigors of Rules 702 and 703 ..., Rule 403 is an unlikely basis for exclusion.").

My colleagues of the concurrence and the majority, by one path or another, invite judges to decide the reliability and relative merit of conflicting expert opinions.

By ordering judicial review of expert opinion supporting the claims of injured parties, and requiring general scientific acceptance as a condition of admissibility, as this court now does, and by rejecting a mass of compelling expert testimony, as the court did in Brock,47 we substantially affect the rights of claimants and the proof required for their compensation. Only diversity jurisdiction opens the federal courts to these cases, and state law properly controls. Under Texas law, claimants do not suffer these disadvantages.48 There can be no doubt that in Texas courts Mrs. Christophersen would have received her jury trial and the Brocks would have recovered their judgment.49 Reflexive citation to the general proposition that federal law governs evidentiary questions does not settle this issue. The Federal Rules' Enabling Act establishes that federal "rules shall not abridge, enlarge or modify any substantive right." 28 U.S.C.A. Sec. 2072(b) (West Supp.1991). The rules set forth in the per curiam opinion sharply curtail claimants' substantive horizons and thus invade "evidentiary rules ... so bound up with state substantive law that federal courts sitting in Texas should accord [them] the same treatment as state courts in order to give full effect to Texas' substantive policy." Conway v. Chemical Leaman Tank Lines, Inc., 540 F.2d 837, 838 (5th Cir.1976).50

This deference to state substantive policy becomes so much more persuasive where, as here, the contents of the state and federal evidence rules are identical, but the federal rule becomes freighted with burdens that depart from both state and federal jurisprudence. Congress has not ordered these changes in the Federal Rules of Evidence. Nor has the Supreme Court done so. By what possible right do we do so? And now where do we turn for an appreciation of federalism and judicial restraint?

The per curiam is unacceptable. I dissent.

KING, Circuit Judge, with whom, REAVLEY, JOHNSON and WIENER, Circuit Judges, join dissenting:

I concur fully in Judge Reavley's thorough and forceful dissent, and I agree with Chief Judge Clark's careful analysis of Rules 702 and 703. I write separately to emphasize what I perceive to be the message and import of the majority opinion.

The tremendous impact of the majority opinion comes not from the opinion itself, but from the simple fact that it could be written at all in the face of the summary judgment record described with such care by Judge Reavley. The juxtaposition of the majority opinion and the record in this case sends a clear message to the district courts in this circuit. Henceforth, a dispute among qualified experts as to the appropriate scientific methodology or reasoning that an expert in the particular field should use to connect the facts to his conclusions is to be resolved by the district judge. If the district judge resolves this aspect of the battle among the experts by concluding that the plaintiff's expert has employed a methodology or reasoning which has not gained general acceptance in the scientific community, the case is to be terminated before trial. The jury will no longer be the arbiter of disputes over alternative forms of scientific methodology or reasoning.

Lest anyone misunderstand, at root this is not a case about the Federal Rules of Evidence, albeit that two of them may have been mangled in the process. It is instead about the outcomes in toxic tort cases.1 What the majority is saying is that in its view, the outcomes in toxic tort cases that have been left to the jury have too often been unacceptable. The majority's solution is to re-write the Federal Rules of Evidence, which have been on the books for more than fifteen years and which, implicitly, have too often been misapplied by "our able trial colleagues."

In terms of its practical impact, today's decision will require a penurious plaintiff represented by a personal injury lawyer on a contingent fee basis, already limited in investigative and expert resources, to pass a preliminary mini-trial on his expert's methodology. At such a mini-trial, the plaintiff will be exposed to all of the industry's heavy artillery, and may be at the receiving end of a district court ruling that will eviscerate his case. To compound the injury, because the majority has conveniently used the hook of the Federal Rules of Evidence, rather than openly announcing a new principle of federal substantive law,2 the district court's ruling will be reviewed under the manifestly erroneous standard. Plaintiffs attempting to establish liability in areas in which scientific methodology or reasoning is not yet well-established will face a nearly insurmountable burden. The result will be to deprive plaintiffs with possibly meritorious claims of a jury's assessment of their right to recovery.

As Judge Reavley and Chief Judge Clark make abundantly clear, the majority disregards the plain language of Rules 702 and 703 in its zeal to approve the district court's dismissal of the plaintiff's evidence in this case. Based on an intellectual elitism that prefers to entrust scientific evaluation to a judge, trained in law, rather than to the good sense of the community, the majority judicially amends the Federal Rules of Evidence in order to deprive the plaintiff of his or her day in court. Before today's decision, I should have thought that a dispute among qualified experts would preclude, not mandate, summary judgment.

The majority, so willing to decry judicial activism in other contexts such as prison overcrowding or supervision over schools, engages in the worst sort of judicial legislation. Undoubtedly, modern toxic tort litigation presents novel, challenging problems for the courts. Such cases may encompass massive numbers of plaintiffs, may involve scientific disputes which the courts are sometimes ill-suited to resolve, and may cause severe and perhaps unwise economic disruptions. If the Federal Rules of Evidence are inadequate in this context, however, then Congress, who enacted them, is the proper branch to amend or repeal them. If a federal substantive rule needs to be devised to supplant state law criteria for recovery, then Congress is the proper branch to devise it.

But at least until now, the decisions when and under what circumstances corporate America or its insurers pay in the area of toxic torts have not been vested by the Constitution or the Supreme Court in federal judges. Those decisions have been determined by state law, as applied by juries. Today's decision places a far greater hurdle in the path of toxic tort plaintiffs in federal court than such plaintiffs face in a state forum. This lack of congruence with the state courts in cases dependent on our diversity jurisdiction is fundamentally at odds with the Supreme Court's decision in Erie. As a result, we encourage reverse forum shopping and deprive plaintiffs of the substantive rights that should be available to them under state law.

The majority, in its haste to achieve more acceptable results, tramples on fundamental Constitutional principles that one might have expected this court, above all others, to hold dear--the separation of powers between Congress and the federal courts, the right to a jury trial and the role of the states in our federal system. With the mandate of neither Congress nor the Supreme Court, we today create federal substantive law, through the technique of amending the Federal Rules of Evidence, in order to tilt toxic tort litigation in favor of defendants. Such result-oriented decision-making can only erode respect for the federal judiciary.