Appellant Sheila Harold appeals from a judgment entered on a jury verdict by the District Court for the Eastern District of Missouri finding in favor of Dr. Kenneth D. Corwin and his public corporation (appellees), and against Harold on her damages claim for medical malpractice.

The issues presented are whether: (1) the voir dire examination conducted by the district court was unduly and unreasonably restricted; and (2) the district court committed prejudicial error by reading to the jury a dictionary definition of a medical phrase contained in an exhibit. We reverse and remand for a new trial.

In July 1982 appellee surgically enlarged Harold's breasts with a silicon implant. At the time of the surgery Harold was a Missouri resident; however, shortly thereafter she moved to Dallas, Texas. After noticing a mass in her right breast, in August 1983 Harold returned to St. Louis to confer with appellee about her breast. Appellee discovered, upon examining Harold's breast, that scar tissue had formed around the silicon implant. Appellee treated the scar tissue by performing a closed capsulotomy: a manual hand procedure that involves squeezing and crushing the scar tissue. After his examination and treatment, and without any further action, testing, or recommendations, appellee informed Harold that the mass in her breast was benign.

Thereafter Harold returned to Dallas, and took no further action until in December 1983 when she once again felt pain in her right breast. After several consultations with Dallas physicians, Harold was admitted to a local hospital. Following her admission, Harold's right breast was removed because a malignant tumor mass had infiltrated the silicon breast implant. Additionally, several auxiliary lymph nodes found under Harold's right arm tested positive for malignant cells. Thereafter, Harold sued appellee alleging negligence in failing to timely investigate the cancerous tumor mass in her right breast on August 26, 1983.

Pursuant to Fed.R.Civ.P. 47(a)1 the district court announced that it would conduct the voir dire examination.2 Thereafter, as permitted by Rule 47(a), Harold submitted a written voir dire request for supplemental inquiries to the veniremen by the court. The district court responded that the requested questions would be reviewed upon completion of the voir dire to see what subject matters contained in Harold's inquiries may not have been touched upon. At the conclusion of the voir dire, the district court refused Harold's renewed request to ask the supplemental questions.

Because of our disposition of the present case, we do not decide whether the district court's voir dire was sufficient. However, because some of the present complaints may recur upon retrial a few observations may be in order. The trial court undoubtedly has considerable discretion in deciding what questions may be put to jurors on voir dire, both in ruling on objections if the questioning is by the attorneys or by refusing to put additional questions requested by a party. 9 C. Wright & A. Miller, Federal Practice and Procedure: Civil Sec. 2482 (1971). This discretion, however, is not without limits. The court, in the realization that the purpose of the voir dire is to afford the parties a trial by a qualified, unbiased, and impartial jury, should at all times be on guard in its questioning in order to assist counsel in the exercise of his or her preemptory challenges and challenges for cause to eliminate those persons with an interest or bias. See Labbee v. Roadway Express, Inc., 469 F.2d 169 (8th Cir.1972) (Voir dire examination is subject to right of parties to have an impartial jury and the questioning must in general attempt to preserve that right). We have examined some of the requested questions submitted by counsel and find them not to be improper; however, on remand it will be left to the district court to re-examine the questions asked of the venirepersons to see if the subject matter of the questions requested by counsel has been fairly and reasonably presented to the venire panel.

Finally, we consider Harold's argument that her case was prejudicially disadvantaged because the district court, sua sponte and over the objections of both parties, read from a dictionary, which was not in evidence, the definition of the word "differentiated." After the submission of the case to the jury for its consideration and deliberations, a question arose from the jurors as to the meaning of the phrase "poorly differentiated tumor." Thereafter, the jury submitted a written question to the court asking for a "layman's" explanation of the phrase "poorly differentiated tumor." The record shows that the phrase had been taken from one of appellee's exhibits that was in evidence, but the phrase had not been either read or explained to the jury, and had been referred to only in questions by appellee's counsel regarding a "lite scan."

As a threshold matter, we make it clear that we find no error in the fact that the exhibit, which was in evidence, was permitted by the district court, with or without either counsel's approval, to go with the jurors into the jury room. This is a highly discretionary decision and will be reversed only for an abuse of discretion. No citations are needed to support this proposition. In contrast, however, exhibits or materials neither in evidence nor used or exhibited before the jury during the conduct of the trial that reasonably can be said to improperly influence the verdict have no place in the jury room. Zagarri v. Nichols, 429 S.W.2d 758, 761 (Mo.1968).

Here the district court, by the use of a dictionary definition of the meaning of the word "differentiated," abortively attempted to do that which none of the medical experts or doctors had done. If the exhibit had been presented to any of the doctors, the phrase "poorly differentiated tumor" could easily have been explained to the jury. "Differentiated," as defined by the court means:

Having a different character or function from the surrounding structure or from the original type; said of tissues, cells, or portions of the cytoplasm.

Before reading this definition to the jury, the district judge himself expressed concern that he was in a strange position, i.e., whether he was "going to cross the lawyers or the jury." He informed the jury that he was confronted with a situation which he had not had in seven years. He told the jury that "poorly differentiated tumor " could not be found as a phrase in the medical dictionary, and that all he could find was "differentiated," and that it was "using elastic " to read the definition that was about to be read. [Emphasis added].

First, there is no question that the definition read was hearsay, i.e., offered to prove or to shed light on the meaning of the phrase "poorly differentiated tumor." Secondly, the meaning given was misleading; i.e., it made it sound as if the phrase meant the structure or masses could be identified as being different by physical palpation, when in fact the phrase "poorly differentiated tumor " is a pathology term describing the microscopic appearance of tumor cells as compared with normal tissue cells. We reject appellee's argument that what was done in the present case was a mere comment upon the evidence. The transgression here is far more serious and strikes at the root of our whole adversary system; i.e., that trials are to be decided based upon only the evidence and the testimony legally admitted under the guidance and supervision of the court during the course of the trial.

Even though we do not believe that the use of a dictionary is prejudicial per se, United States v. Steele, 785 F.2d 743, 746 (9th Cir.1986) (citing United States v. Bagnariol, 665 F.2d 877 (9th Cir.), cert. denied, 456 U.S. 962, 102 S. Ct. 2040, 72 L. Ed. 2d 487 (1982)), we are constrained to disapprove what was done here because of the failure of the district court to take the proper safeguards in its attempt to define the phrase. United States v. Griffin, 756 F.2d 1244, 1250-52 (6th Cir.), cert. denied, 474 U.S. 837, 106 S. Ct. 114, 88 L. Ed. 2d 93 (1985).

In United States v. Birges, 723 F.2d 666 (9th Cir.), cert. denied, 466 U.S. 943, 104 S. Ct. 1926, 80 L. Ed. 2d 472 (1984) (citing United States v. Gunter, 546 F.2d 861, 869 (10th Cir.1976), cert. denied, 431 U.S. 920, 97 S. Ct. 2189, 53 L. Ed. 2d 232 (1977)), the court noted:

Questions or disputes as to the meaning of terms which arise during jury deliberations should be settled by the court after consultation with counsel, in supplemental instructions. Such guidance will avoid the danger that jurors will use the dictionary to construct their own definition of legal terms which do not accurately or fairly reflect applicable law.

United States v. Birges, 723 F.2d at 670-71.

Granted Birges and Gunter are criminal cases, this, however, in no way detracts from the announced salutary rule given therein. Here the district court's actions suffer from three vices: the definition was read from the dictionary, was read to the jury over counsel's objections without any supplemental instructions or directions by the district court, and the definition given neither fairly nor accurately described and phrase "poorly differentiated tumor " in the medical context in which the term was used. In fact, no safeguards were taken other than the court's own announced perplexity with a novel situation.

Under the circumstances shown herein, we find that the prejudice to Harold's case is apparent and that the measures taken by the court in reading from the dictionary caused Harold substantial prejudice. See Osborne v. United States, 351 F.2d 111, 119 (8th Cir.1955).

Accordingly, the judgment is reversed and remanded for a new trial.

LAY, Chief Judge, concurring.

I am pleased to join Judge McMillian's well-reasoned opinion. I write separately, however, because of counsel's suggestion that the trial court erred in foreclosing counsel's personal participation in the voir dire.

Fed.R.Civ.P. 47(a) provides the trial court the discretion to permit counsel to conduct voir dire or to allow counsel and the court both to conduct voir dire. My experience is that most trial judges, both federal and state, allow counsel to participate in the voir dire. On the other hand I fully realize that some federal judges feel that counsel should not participate in the voir dire primarily because (1) counsel will attempt to prejudice potential jurors and (2) reasonable time limitations will be violated by lawyers abusing the right to question the panel. It seems to me the middle ground of allowing both counsel and court to participate in the voir dire is the most fair and flexible means to provide a fair selection of jurors. Furthermore, joint participation more readily conforms to the flexibility and contemplation of rule 47(a).

My colleagues in Hicks v. Mickelson, 835 F.2d 721 (8th Cir.1987), have suggested otherwise. Their "additional comments" were not necessary to that opinion anymore than my present comments are necessary to this opinion. I do, however, disagree with them. In thirty-eight years as a lawyer and occasional trial judge I have participated in literally hundreds of voir dire of juries. I say this not to suggest that my judgment is any better than theirs. They simply speak from their experience and I speak from my own. We are all products of our experiences. However, aside from our own personal backgrounds, we are all governed by a rule of procedure which does not provide an inflexible view urged by the panel in Hicks.

Rule 47(a) was drafted in contemplation of lawyer participation; its origin is not of singular experience, but represents an empirical view of distinguished lawyers and judges who constituted the Advisory Committee on the Federal Rules of Civil Procedure. These rules were recommended by the Judicial Conference of the United States and approved by the Supreme Court. Although rule 47(a) allows the court to conduct voir dire it certainly does not, however, preclude participation by the trial lawyer, as my colleagues suggest the better policy to be. Rather, rule 47(a) unequivocally states that trial judges may either permit attorneys to conduct the examination of prospective jurors or they may conduct the examination themselves, allowing counsel to jointly participate by propounding additional questions for them to ask.

The grave danger of a voir dire controlled solely by the judge is found in the unnecessary reversal of cases where the judge offers allegedly neutral, flat and non-penetrating questions to potential jurors.1 Where the judge alone conducts the voir dire it generally provides a paucity of information to allow a judgmental exercise for peremptory challenges by counsel. Furthermore, if lawyers are continually denied a fair participation in voir dire, inflexible legislation requiring voir dire solely by lawyers may well be the answer. See S. 677, 98th Cong., 1st Sess., 129 Cong.Rec. S2122 (1983) (amending Fed.R.Civ.P. 47(a) to require that each side's counsel be afforded at least thirty minutes to question prospective jurors). Voter registration lists contain a far greater number of trial lawyers than trial judges.

Any busy trial judge must candidly admit that he or she knows far less of a given case at the time voir dire commences than the lawyers who have prepared the case for months or years. This reason alone compels the conclusion that lawyers should participate in voir dire. Notwithstanding this basic concern, I fail to see the logic of the reasons given to exclude the trial lawyer from the most important phase of the adversary process.

It is argued that many trial lawyers are not skilled in voir dire of juries. I intend no disrespect, but neither are many trial judges. Nevertheless, if the lawyer persists in asking the same repetitive questions to each juror, the lawyer can be directed by the court to limit specific questions to individual jurors. If, however, the trial judge persists in asking innocuous questions there is little the lawyer can do.

As the Hicks panel points out, some lawyers also foolishly claim that effective voir dire can win cases by "prejudicing juries." Such ill-strategy provides a denigrating view of the jury system. My experience has been that most lawyers who set out with such strategy meet resentment from the composite intelligence of the average jury. Although this professed strategy may provide a great display of vanity at trial seminars, most skilled lawyers will advise against it. Voir dire is necessary and most effective when it seeks to disclose latent prejudice and bias of potential jurors. This critical information is seldom produced in judge-offered interrogatories.

It is also argued that allowing lawyers to conduct voir dire will abuse the time limitations inherent in a trial. This also is unrealistic. If a trial judge concludes that a lawyer is abusing the process by either "prejudicing" the jury or abusing its time limitations, the judge can effectuate reasonable rules of procedure to curtail the abuse. Proper and experienced judicial oversight is exercised continually in the course of a trial. The court provides reasonable control in discovery, in opening statements, excessive and repetitive direct examination, abusive cross-examination and in limitation of content and time of closing arguments. The conduct of voir dire is no different.

The rigid, inflexible control suggested by my colleagues in Hicks and adopted by many federal trial judges is to have the lawyers submit written questions for the court to propound to the jury. My trial experience compels me to conclude that this is akin to taking a bath with your clothes on or listening to the London Philharmonic play Beethoven's Fifth with ear plugs in the ear. I think most lawyers will agree that this is mere rhetorical exercise and totally ineffective. The neutral response from the prospective juror does little to inform the lawyer as to the possibility of latent prejudice. These responses are a far cry from the spontaneity of the give and take between lawyers and jurors in an informal, relaxed, and probative discussion of the subject matter.2 It seems to me that rule 47(a) contemplates more than a rigid, innocuous "please raise your hand if it applies to you" process.

The critical importance of a truly fair and impartial jury is essential to due process in the adversary system and basic to our conception of jurisprudence. See Alderman v. Austin, 695 F.2d 124, 130 (5th Cir.1983) (Fay, J., dissenting) ("That the voir dire process is vital to the fairness of jury trials is beyond cavil.") (citing United States v. Peterson, 483 F.2d 1222, 1226 (D.C.Cir.1973)). Voir dire was devised to insure to the fullest extent possible that only a fair and impartial jury will be selected. Effective voir dire examination can develop a prospective juror's state of mind and reveal many subtle factors which can influence the juror, many of which are seldom brought out when lawyers are precluded from participating in the examination of the venire. This not only enables the court to determine actual bias, but also affords counsel the essential facts necessary to intuitively exercise his or her peremptory challenges. Photostat Corporation v. Ball, 338 F.2d 783, 786 (10th Cir.1964). The permissive nature of rule 47(a) is one of common sense and I would hope more trial judges would adhere to it. I believe this would tend to reduce claims of error such as those that occurred here.

Upon remand I would hope the trial judge, in the exercise of his discretion under rule 47(a), would allow the lawyer to participate in the conduct of the voir dire examination.