Plaintiffs Paula Renfroe and Nancy Deitchman have brought pharmaceutical products liability actions in the United States District Court for the Eastern District of Missouri, Eastern Division, against E.R. Squibb & Sons, Inc. (Squibb) and other drug companies for injuries allegedly caused by in utero exposure to the drug diethylstilbestrol (DES). Paula Renfroe, born in 1964, has an adenocarcinoma of the cervix; Nancy Deitchman, born in 1956, has an oat cell carcinoma in the endometrial lining of her uterus.

In connection with this litigation, Squibb in February 1983 asked the United States District Court for the Northern District of Illinois, Eastern Division, to issue subpoenas for deposition which required, in essence, the production of every document in the Registry for Hormonal Transplacental Carcinogenesis (the Registry) at the University of Chicago. The Registry monitors the clinical, pathological, and epidemiological aspects of clear cell adenocarcinoma of the genital tract and is the only centralized repository of data on that disease. On March 4, 1983, Dr. Arthur L. Herbst, custodian of the Registry's records and chairman of the University of Chicago's Department of Obstetrics and Gynecology, filed a motion to quash the deposition subpoenas, pursuant to Fed.R.Civ.P. 45(b), claiming that the Registry data were privileged and confidential. Squibb opposed Dr. Herbst's motion.

On March 23, 1983, a United States District Judge entered orders, with an accompanying Memorandum Opinion, granting in major part Dr. Herbst's motion to quash. Insofar as Squibb's subpoena requested records relating to plaintiffs and their mothers, however, the court ordered Dr. Herbst to disclose any records or other information which pertained solely to them because, in their cases, the need for confidentiality was lost when this litigation began.

Squibb thereafter filed a Motion to Reopen for Reconsideration and for Relief, pursuant to Fed.R.Civ.P. 59(e); and a Motion to Vacate, to Reopen and for Relief, pursuant to Fed.R.Civ.P. 60(b)(2). The court entered orders denying these post-judgment motions.

Squibb appeals from the orders granting Dr. Herbst's motion to quash the deposition subpoena and from the orders denying its motion to reopen and reconsider. For the reasons set out below, we vacate each of the district court's orders and we remand for further proceedings.

* Dr. Herbst, the appellee before us, is not a party to any case and will not testify at the trials. His connection with the cases will appear below. He was on the faculty of the Harvard Medical School and on the staff of Massachusetts General Hospital when he began in 1969 a study of clear cell adenocarcinoma of the vagina of young women. In 1971, he and his colleagues became the first medical researchers to suggest an association between exposure to DES in the womb and the development of clear cell adenocarcinoma of the vagina or cervix at an early age. See Herbst, Ulfelder & Poskanzer, Adenocarcinoma of the Vagina--Association of Maternal Stilbestrol Therapy with Tumor Appearance in Young Women, 284 New Eng.J.Med. 871 (1971). All parties agree that the 1971 study is of enormous significance in this field of medicine and that Squibb will in all likelihood be confronted with it in one form or another at trial on the merits, even though Dr. Herbst himself will not testify.

In order to monitor the clinical, pathological, and epidemiological aspects of clear cell adenocarcinoma of the genital tract (the disease), in 1972 Dr. Herbst set up the Registry. Among the Registry's initial goals were the determination of the incidence of the disease, the best means of treatment, the relationship between prenatal hormone treatment and the occurrence of the disease, the other factors which promoted the disease, and the other sites of the disease in the genital tract. Since the Registry's beginning, it has received over 500 case files, and Dr. Herbst and his colleagues have published more than a dozen articles which report all significant findings from the Registry on the subject of DES and clear cell adenocarcinoma.

To establish the Registry, Dr. Herbst sent letters of inquiry to all departments of medical schools in the United States, to hospitals throughout the world that specialize in cancer treatment, and to the World Health Organization. He also placed announcements in various major medical journals. In his solicitations, Dr. Herbst requested records of all women born after 1940 who had contracted clear cell adenocarcinoma of the genital tract; he promised that the Registry would keep confidential all information submitted to it. After he received records for a particular case, Dr. Herbst attempted to contact the patient, and where possible, her mother's obstetrician, to obtain additional records and information about the mother's possible use of DES. He requested this information with an additional written promise of confidentiality.

Dr. Herbst continually updates and annually requests information from treating physicians on the status of their patients' diseases, the treatment being used, and any complications from the treatment. He also continues to solicit information on newly reported cases of clear cell adenocarcinoma of the female genital tract and on newly reported cancers in women exposed prenatally to other sex hormones. Because the FDA did not proscribe the use of DES for pregnant women until 1971, DES-associated cases of clear cell adenocarcinoma are expected to continue to appear until well into the 1990's. Continued monitoring through the Registry, therefore, will be vital in order to learn more about the disease and how to treat it.

When a discovery subpoena is issued against a party or person who objects to the disclosure of information called for, that party or person can move for either a protective order pursuant to Fed.R.Civ.P. 26(c), or an order quashing or modifying the subpoena, pursuant to Fed.R.Civ.P. 45(b). Fed.R.Civ.P. 45(d)(1). Dr. Herbst has moved for such protection, arguing not only that Squibb has no need to see Registry data, but also that the subpoena as issued is unreasonable and oppressive.

Generally, the Federal Rules of Civil Procedure allow for broad discovery: "Parties may obtain discovery regarding any matter, not privileged, which is relevant to the subject matter involved in the pending action * * *." Fed.R.Civ.P. 26(b)(1) (emphasis added). When protection from a deposition subpoena is sought, the court "must apply a balancing test to determine whether the need of the party seeking disclosure outweighs the adverse effect such disclosure would have on the policies underlying the [claimed] privilege." Equal Employment Opportunity Commission v. University of Notre Dame du Lac, 715 F.2d 331, 338 (7th Cir.1983). That is, the court must compare "the hardship to the party [or person] against whom discovery is sought, if discovery is allowed, with the hardship to the party seeking discovery if discovery is denied." Marrese v. American Academy of Orthopaedic Surgeons, 726 F.2d 1150, 1159 (7th Cir.1984) (en banc ). Thus, the nature and magnitude of Dr. Herbst's, the Registry's, and Squibb's competing hardships must be reviewed. We turn first to the hardship a subpoena will cause to Dr. Herbst and the Registry.

* Dr. Herbst vigorously argues that a subpoena will literally destroy the Registry if it pierces the Registry's promised confidentiality. He buttresses his argument with affidavits from a number of experts in the field of medical science and epidemiology: Dr. Robert E. Scully, Professor of Pathology, Harvard Medical School, and a member of the Pathology Office of the Registry; Dr. Leonard T. Kurland, Chairman and Professor of Epidemiology, Mayo Medical School and Professor of Epidemiology, School of Public Health, University of Minnesota; Dr. William M. Halnzal, Professor of Epidemiology, University of Illinois School of Public Health and Senior Epidemiologist, Illinois Cancer Council; Dr. Philip Cole, Professor and Head of the Department of Epidemiology, University of Alabama School of Public Health and Associate Director of that university's Comprehensive Cancer Center; Dr. Kenneth Noller, Principal Investigator of the Mayo Clinic DES Examination Center; Dr. Raymond H. Kaufman, Professor and Chairman of the Department of Obstetrics and Gynecology and Professor of Pathology, Baylor College of Medicine, Houston; and Dr. Hugh R. Barber, a gynecologist in private practice in New York City. Drs. Scully, Noller, Kaufman and Barber specifically averred that they had in the past submitted information and other material to the Registry based on Dr. Herbst's representations of strict confidentiality and that this cooperation would cease should it become apparent that such material was amenable to subpoena.

Dr. Herbst and his affiants exhort us on the profound societal impact a subpoena here will have. They lecture us on the importance of confidentiality in epidemiological research, the significance of the Registry as a unique and valuable research project, the necessity for its continuation, and the harm that Squibb's discovery of the data in the Registry could cause, not only to the Registry, but to other epidemiological research projects as well.

Squibb, not surprisingly, makes absolutely no counter-showing. It appears to concede that the loss of confidentiality here will adversely affect the Registry and that, as the district judge stated, "all society will be the poorer * * * [and] a unique and vital resource for learning about the incidence, causes and treatment of adenocarcinoma will be lost."

The issue of confidentiality thus appears to us not an insurmountable one. It is a familiar problem in discovery cases and measures to preserve it are easily contrived, especially if one does not confine the protection to mere deletion of informants' names from copies of papers furnished. Squibb is willing to bear any expense that discovery of the papers may require. The subpoena is of course too broad as issued, but there is nothing unusual about that. Such a sweeping subpoena means about as much as the asking price for a rug in an Oriental bazaar. It is normally just a means of opening discussion between discoverer and discoveree. The discoverer asks for too much because he is not, until he is told, aware of the discoveree's problems. When a court is confronted with a motion to quash such a subpoena, its duty is not to deny any discovery, but to reduce the demand to what is reasonable, considering the discoverer's needs and the discoveree's problems.

Dr. Herbst also asserts a second reason to quash the subpoena here, namely, that the forced premature disclosure of research results adversely affects first amendment rights to academic freedom and freedom of the press and has a "chilling" effect on the American academic community. Rephrasing Dr. Herbst's position, it is that despite all the talk about confidentiality, the fullest possible safeguards to confidentiality would not mitigate his resistance to the subpoena in the slightest. His real and deepest objection is that he must be allowed to divulge to the public the results of his studies only in his own time and way. His conclusions are still tentative, data are still being collected. He has not yet submitted his case to "peer review" as is normal in the scientific community, despite his many publications.

This circuit has expressly recognized that such an effect of premature publicity must be guarded against, in Dow Chemical Co. v. Allen, 672 F.2d 1262, 1276 (7th Cir.1982):

[T]here are several reasons to think [subpoenas] * * * capable of chilling the exercise of academic freedom. * * * [E]nforcement of the subpoenas would leave the researchers with the knowledge throughout continuation of their studies that the fruits of their labors had been appropriated by and were being scrutinized by a not-unbiased third party whose interests were arguably antithetical to theirs. It is not difficult to imagine that that realization might well be both unnerving and discouraging. * * * In addition, the researchers could reasonably fear that additional demands for disclosure would be made in the future. * * * To these factors must be added the knowledge of the researchers that even inadvertent disclosure of the subpoenaed data could jeopardize both the studies and their careers.

Thus, indisputably, there is a vital public interest in promoting research of the type the Registry carries out. We agree arguendo that the Registry files may enjoy a qualified privilege and are not to be pried into unnecessarily; however, such privileges are not absolute. They must yield if to enforce them would produce a miscarriage of justice. See, e.g., Branzburg v. Hayes, 408 U.S. 665, 709-710, 92 S. Ct. 2646, 2670-2671, 33 L. Ed. 2d 626 (1972) (Powell, J., concurring); Dow Chemical Co. v. Allen, supra. Because the party seeking discovery may show an absolute necessity to allow discovery, we must next consider the hardship to Squibb should we uphold the district judge's grant of Dr. Herbst's Motion to Quash.

Squibb argues that without the sought after information, it suffers hardship because it is unable to adequately defend itself at trial on the merits. For example, Squibb has asserted throughout these proceedings that "owing to Dr. Herbst's preeminence and dominance of [his DES] research, his published work and/or articles have universally been relied upon by plaintiffs' experts in DES litigation," [Joint Appendix (JA) AA254]; that "Dr. Herbst's published studies have been used over and over again * * * to show the relationship between DES-exposed and nonexposed women as the basis for plaintiff's causation argument," [JA AA262]; that without access to Registry data, Squibb "is put in a position of being unable to defend the basic issue raised in the DES litigation (i.e., causation) adequately, given the unavailability of the underlying data upon which Dr. Herbst's conclusions are based," [JA AA272]; that Dr. Herbst and the Registry "are the 'sole monitors' * * * responsible for investigating a possible association between prenatal DES exposure and clear cell adenocarcinoma of the vagina * * *," [JA AA273]; that by publishing articles based on Registry data, Dr. Herbst "has become an integral part of the entire causation element of [DES] cases," [JA AA281]; and that the plaintiffs' cases were brought "based upon Dr. Herbst's publications in the medical community" based on Registry data [JA AA282]. These we do not think are overstatements.

The district court recognized that Squibb needs access to Registry data "in order to make cross-examination meaningful," yet it doubted that Squibb's need for the material is "truly compelling." Several of the fact-findings leading to this latter conclusion are subject to challenge, however.

For example, the court found that plaintiffs' experts rely solely on Dr. Herbst's original 1971 article, an article which he published before he founded the Registry. Thus, according to the court, although the data in the Registry might tend either to corroborate or contradict Dr. Herbst's 1971 study, plaintiffs would not likely confront Squibb either directly or indirectly with Registry data. That plaintiffs' experts rely on the cumulative information contained in all of the articles which Dr. Herbst has published since 1971, however, is clear; Squibb cites numerous examples. To find otherwise is "clear error."

The court also found that Squibb had a lessened need for the data because Dr. Herbst himself was not testifying at the trial on the merits. Yet, it is undisputed that Dr. Herbst's published articles will be used against Squibb, and that Squibb needs access to the underlying data to prepare its defense.

What Squibb is threatened with is having Dr. Herbst as a potent expert witness against it without his ever taking the stand or being subject to cross-examination. The situation is unique because there is placed in the hands of a nonwitness the capability of influencing the verdict far beyond that enjoyed normally by the most qualified expert witness who does take the stand. It appears likely that any expert who challenges Dr. Herbst's views can be refuted by the argument that this witness does not know the most pertinent evidence, the contents of the Registry files, and Dr. Herbst does know that.

Certainly, Squibb is entitled to cross-examine plaintiffs' experts on the data underlying their opinions regarding DES. Fed.R.Evid. 703, 705. Cross-examination is a fundamental right that a court should abridge only to curb abuse. Alford v. United States, 282 U.S. 687, 691, 51 S. Ct. 218, 219, 75 L. Ed. 624 (1931); Union Automobile Indemnity Association v. Capitol Indemnity Insurance Co., 310 F.2d 318, 321 (7th Cir.1962). Although the scope of cross-examination is left to the discretion of the trial judge, where the circumstances are such that a restriction on cross-examination is so severe as to constitute a denial of the right, the particular restriction in question may constitute an abuse of discretion. See Reilly v. Pinkus, 338 U.S. 269, 275, 70 S. Ct. 110, 113, 94 L. Ed. 63 (1949) ("It certainly is illogical, if not actually unfair, to permit witnesses to give expert opinions based on book knowledge, and then deprive the party challenging such evidence of all opportunity to interrogate them about divergent opinions expressed in other reputable books." Id.). The denial of discovery here has effectively precluded Squibb from engaging in any meaningful cross-examination of plaintiffs' experts' opinions which are based, it appears, exclusively on Dr. Herbst's publications.

Finally, and most important to the decision to quash the subpoena, was that Squibb made no showing as to what it hoped to prove with the Registry data. The court stated that "Squibb's asserted 'need' for this material is based, on the present record, on nothing more than a speculative hope that it will find something that undermines the widely-accepted conclusions of the medical profession about DES," and that "to undermine Dr. Herbst's published findings, Squibb must, it appears, demonstrate that Dr. Herbst materially misrepresented the data in the Registry in his writings," speculation which the court could not readily permit about a "man of Dr. Herbst's stature in his profession." An unfriendly critic might translate this to mean that Dr. Herbst is infallible and inquiry into the bases of his opinions would be futile no matter how much aided by discovery. Only if Dr. Herbst can be detected knowingly making false representations would inquiry facilitate the administration of justice; that he could be honestly mistaken is not to be thought of.

Squibb, however, makes absolutely no allegation in the record of any impropriety on Dr. Herbst's account. It merely states that its discovery request is only an attempt "to gather discoverable and relevant information as to the methodology used by Dr. Herbst and the Registry in correlating their statistical information." In particular, Squibb expresses its concern "with the credibility, accuracy and validity of those statistics and conclusions * * *," and it contends that it must obtain access to the data underlying those findings so that it can "critically scrutinize" them and "effectively and efficiently cross-examine" plaintiffs' experts. Squibb relies on Wright v. Jeep Corp., 547 F. Supp. 871, 874 (E.D.Mich.1982), a case somewhat analogous to the present action, where the court stated:

Researchers accumulate basic data from an infinite number of sources. Sometimes it has been acquired by others and is used by the researcher. Sometimes it is acquired specifically for this research project. Sometimes and usually, both sources are used. The data are analyzed, compared and contrasted and from it all, the researcher draws conclusions using his analysis to document and support his report which in turn supports his conclusions. The value of the conclusions turns on the quality of the data and the methods used by the researcher in his analysis of that data as well as the skill and perception of the researcher.

So if the conclusions or end product of a research effort is to be fairly tested, the underlying data must be available to others equally skilled and perceptive.

In the eyes of the law there is no such thing as an infallible witness. There are only some who are more likely to be correct than others. We can assume the district judge might, without abuse of discretion, regard Dr. Herbst as one of those who are more rather than less likely to be correct, and to put willful falsification aside in all events. The discovery well may be, perhaps will be, as futile as the judge supposes. The trouble is, when discovery has not been tried, no one can say for sure whether it is going to be futile or not. The expectation that it will be futile is, therefore, not the certainty that justifies cutting off a party's discovery rights without any effort to satisfy them even in the most essential particulars.

Importantly, a study of this sort may have a number of different, but inadvertent, biases present. See, e.g., Sackett, Bias in Analytic Research, 32 J. Chronic Disease 51-63 (1979). Squibb, for example, has presented evidence that not all physicians in the medical community report all their clear cell adenocarcinoma cases to the Registry. This fact could very well make the statistical basis for Dr. Herbst's published conclusions inaccurate or incomplete. Moreover, although Dr. Herbst regularly summarizes his cases and comments on his findings, no one has yet publicly reviewed the actual core data indicating the basis and evidence upon which he classifies a patient's "exposure" or "nonexposure" to DES.

Dr. Herbst makes but a weak attempt to contradict Squibb's professed need for Registry data. He argues that "although Squibb asserts it cannot adequately defend itself against such [DES] suits without the Registry data, it has done so to date with a fair success ratio." A "fair success ratio" in litigation, however, bears little correlation to Squibb's need for the information it seeks. The fact that with the assistance of able counsel Squibb has won a few of the DES cases which have gone to trial is insufficient reason to continue to subject Squibb to the risk inherent in a jury verdict without proper means of defense. It could easily be that Squibb would be hit with large verdicts on the basis of conclusions that are avowedly only tentative. After a series of final judgments, Dr. Herbst might one day announce that new information has led him to abandon his previous conclusions. For Squibb to prepare properly a defense on the causation issue, access to the Registry data to analyze its accuracy and methodology is absolutely essential.

* To weigh competing hardships to determine the appropriateness of discovery is clearly within the responsibility of the trial judge, in the first instance. Our review authority here, therefore, is quite limited. It is well-established that a court of appeals will reverse a district court's order quashing a subpoena duces tecum only where there is a clear abuse of discretion. Chrysler Corp. v. Fedders Corp., 643 F.2d 1229, 1240 (6th Cir.), cert. denied, 454 U.S. 893, 102 S. Ct. 388, 70 L. Ed. 2d 207 (1981); Marine Petroleum Co. v. Champlin Petroleum Co., 641 F.2d 984, 991 (D.C.Cir.1980); Voegeli v. Lewis, 568 F.2d 89, 96 (8th Cir.1977); Silkwood v. Kerr-McGee Corp., 563 F.2d 433, 436 (10th Cir.1977); Keyes v. Lenoir Rhyne College, 552 F.2d 579, 581 (4th Cir.), cert. denied, 434 U.S. 904, 98 S. Ct. 300, 54 L. Ed. 2d 190 (1977); Imperial Ethiopian Government v. Baruch-Foster Corp., 535 F.2d 334, 337 n. 8 (5th Cir.1976); Premium Service Corp. v. Sperry & Hutchinson Co., 511 F.2d 225, 229 (9th Cir.1975); Fifth Avenue Peace Parade Committee v. Gray, 480 F.2d 326, 333 (2d Cir.1973), cert. denied, 415 U.S. 948, 94 S. Ct. 1469, 39 L. Ed. 2d 563 (1974).

Under the "abuse of discretion" standard of review, the relevant inquiry is not how the reviewing judges would have ruled if they had been considering the case in the first place, but rather, whether any reasonable person could agree with the district court. Washington v. Sherwin Real Estate, Inc., 694 F.2d 1081, 1087 (7th Cir.1982). Compare Marrese v. American Academy of Orthopaedic Surgeons, 726 F.2d 1150, 1159 (7th Cir.1984) (en banc ). The term "abuse of discretion" is perhaps unfortunate in that it sounds more pejorative than the user really intends. That is so here. We are confronted with an able and thoughtful analysis. Yet, a judge abuses his discretion when his decision is based on an erroneous conclusion of law or where the record contains no evidence on which he rationally could have based that decision, Premium Service Corp., 571 F.2d at 229, or where the supposed facts found are clearly erroneous as found. Thus, "if reasonable men could differ as to the propriety of the [district] court's action, no abuse of discretion has been shown." Washington v. Sherwin Real Estate, Inc., 694 F.2d at 1087, quoting Smith v. Widman Trucking & Excavating, Inc., 627 F.2d 792, 796 (7th Cir.1980). This standard of review applies equally to discovery sought in a proceeding ancillary to the principal action. Marine Petroleum, 641 F.2d at 991.

Squibb sought with its subpoena virtually every document within Dr. Herbst's Registry. The subpoena called for:

1. All records and documents, including but not limited to, medical records, history and treatment pertaining to all women accessioned into the Registry with a diagnosis, suspected diagnosis or history of [:] clear-cell adenocarcinoma of the vagina and/or cervix;

2. * * * adenosis of the vagina or cervix;

3. * * * clear-cell adenocarcinoma and adenosis of the vagina or cervix;

4. * * * DES exposure; [or]

5. * * * exposure to synthetic estrogens;

6. All records and documents pertaining to DES, diethylstilbestrol, stilbestrol or other synthetic estrogens;

7. Copies of all questionnaires, forms, sheets or other investigative documents used to determine and/or study the estrogen exposure and history of each clear-cell adenocarcinoma case accessioned into the Registry;

8. Copies of all protocols, study protocols, guidelines and procedures used in connection with the Registry, including the data accumulated therein;

9. All calculations, computer programs and statistical analysis charts, schedules and tests applied to or used in connection with the data accumulated in the Registry; [and]

10. All drafts and final articles, studies or reports prepared or published referring to the Registry data.

Although Squibb has no right to all of the Registry data it calls for, not all the data in the Registry are undiscoverable to Squibb. Because the Federal Rules of Civil Procedure give the trial judge the power to fashion virtually any order "which justice requires to protect a party or person from annoyance, embarrassment, oppression, or undue burden of expense, * * *," Fed.R.Civ.P. 26(c), it surely was possible here to have fashioned a protective order or modified the subpoena in some way so as to protect Dr. Herbst and his Registry against any loss of confidential information and unreasonable financial and temporal costs while still giving Squibb discoverable information at its sole expense. To give Squibb virtually no information we reluctantly call an abuse of discretion.

Even assuming, as Dr. Herbst argues, the redaction of names and addresses from the patient files will not protect a patient's confidentiality in view of the vast amount of data the pharmaceutical industry has amassed and the almost unlimited resources it appears to have available to it, Squibb is still entitled to some discovery. Of course, we do not presume here to be able to fashion a discovery order ourselves. Nevertheless, we can conceive of various types of orders which could sufficiently provide Squibb with at least some information while at the same time protecting the Registry. For example, courts have ordered disclosure of confidential material to an impartial third party or expert, to be paid by the party seeking discovery. E.g., Triangle Manufacturing Co. v. Paramount Bag Manufacturing Co., 35 F.R.D. 540, 543 (S.D.N.Y.1964); Cities Service Oil Co. v. Celanese Corp. of America, 10 F.R.D. 458 (D.Del.1950).

Importantly, Squibb requests two wholly different types of data here. In particular, items 1-5 in Squibb's subpoena ask for data which others have submitted to Dr. Herbst arguably because of his promise of strict confidentiality. Certainly, "compelled production of a researcher's notes may inhibit prospective and actual sources of information, thereby obstructing the flow of information to the researcher, and through him or her, the public." In re Grand Jury Subpoena Dated January 4, 1984, 583 F. Supp. 991 at 994 (E.D.N.Y.1984). Yet, where substantial need is shown, even confidential matter may be revealed. Id. at 2. Although we do not believe that this case calls for release of identifying information, as was sought in the above-cited grand jury case, non-identifying information should be, if possible, provided to Squibb.

As to the information Squibb requests in items 6-10 of its subpoena, on the other hand, an entirely different analysis may be necessary, because this information might not include patient confidences. Moreover, this is information with which Squibb can review Dr. Herbst's statistical methodology. This information is important for Squibb adequately to cross-examine witnesses who rely on Dr. Herbst's studies. We think the discovery of this information ought to be wholly factual. No discovery should be allowed of any material reflecting development of Dr. Herbst's ideas or stating his or others' conclusions not yet published.

It is unclear whether either Squibb, Dr. Herbst, or the court attempted to reach any middle ground with respect to the scope of the discovery order. The nature of Dr. Herbst's real objection being what it is, we doubt if he would voluntarily comply with any discovery whatever. However, some things must be more objectionable to him than others. The discovery should be no more intrusive than is necessary to avoid a miscarriage of justice. Anything not necessary must be viewed as covered by the privilege.

The opinion below, clear and exhaustive as it is, is written mostly on the premise that the issue is all or nothing: the entire subpoena as served, or else no discovery at all. The possibility of a partial solution is not adequately dealt with, though that is always the real issue in cases such as this. Only the possibility of deleting names to protect confidentiality is summarily dealt with by the assertion that such deletions could not be adequate to protect confidentiality. No consideration is given to use of a protective order which would allow information to be disclosed to attorneys and expert witnesses for use in DES litigation only. No consideration is given to the possibility of disclosure to an impartial third party, again limited to use in DES litigation.

We do not undertake to say how the very real problem the case presents may best be solved. We think this is best ascertained by in-chambers negotiation with counsel, where, of course, plaintiffs should be represented too. Only then can the court ascertain where and how seemingly nonnegotiable positions can be called upon to yield a little. The district judge is going to have to place himself in the shoes of the trial judge to be, who will be responsible for getting justice done. He or she would be responsible for this discovery also, except that Squibb must go into another court to enforce its subpoena, an unfortunate happenstance for all concerned.

The court is not to regard this kind of situation as a game or sporting contest, or rely too much on inadequacies in a party's presentation. If a possibility of a miscarriage of justice exists, the reputation and repute of the court itself is placed in jeopardy. The district judge persuaded himself that this possibility was minimal, but did so on the basis of erroneous findings and premature assignment of infallibility as we have specified. Probably there are many Americans who find drug companies in general to be objects they love to hate. This being so, it is the special responsibility of courts to see to it that cases against them at least are fairly presented, and defense not unduly hobbled. At least we should not make matters worse than they already are. It was the same way with railroads a century ago, and no doubt there will be some now unforeseen scapegoat for public ire in A.D. 2084, at which time drug companies will be the objects of commiseration that railroads are today.

Because the Federal Rules "permit the broadest possible scope of discovery and leave it to the enlightened discretion of the district court to decide what restrictions may be necessary in a particular case," Wright & Miller, Federal Practice and Procedure, Civil Sec. 2036 at 268, we remand this case to the district court. We leave it to the district judge to hear the parties and to fashion as inventive an order as the necessities of this unique case dictate, one which allows Squibb the least necessary amount of information to avoid a miscarriage of justice without doing needless harm to Dr. Herbst or his Registry.

VACATED AND REMANDED.