The troublesome legal issue of the compelled disclosure by a journalist of his sources of information gave rise to this interlocutory appeal (28 U.S.C. § 1292(b)) from the District Court. It comes to us in the context of a civil action for libel, as contrasted with the criminal setting in which the Supreme Court has most recently examined the question and sustained compulsion. Branzburg v. Hayes, 408 U.S. 665, 92 S.Ct. 2646, 33 L.Ed.2d 626 (1972). Although there may be discernible degrees of difference in the social interests attaching to the exaction of testimony in the one field as compared with the other, we have concluded, on the basis of both authority and reason, that civil litigation has its entitlements on proper occasion to the pursuit of truth wherever it may be found. We find that this record presents one such occasion; and we affirm the District Court.
I
The complaint in the District Court alleged that on December 14, 1970, the following item was published in a syndicated daily newspaper column known as “The Washington Merry-Go-Round”:
WASHINGTON EXPOSE
Records Stolen? — With the government digging deeper into the financial affairs of the United Mine Workers, the union’s President Tony Boyle and General Counsel Ed Carey spent hours recently going through the records. Later, they were seen removing box-fuls of documents from Boyle’s office. Not long afterward, Carey made an official complaint to Washington police that burglars had struck at union headquarters. Among the goods reported stolen: a boxful of “miscellaneous items.” The Justice Department is investigating.
The complaint further alleged that this item had been written by appellant and a co-defendant, Jack Anderson, with a malicious purpose to damage appellee’s reputation, and with the effect of causing such injury. Substantial compensatory and punitive damages were sought. An answer was filed on behalf of appellant and Anderson, which described appellant as employed by the latter for the purpose of “research, news gathering, and journalistic writing in the course of this employment.” Authorship was admitted of the paragraph upon which the complaint was based, but liability was denied. Extensive discovery proceedings were then engaged in by both sides. The facts which follow are derived from a deposition taken of appellant.
Prior to publication of the item in question, appellant established, by direct inquiry at the police department, that the burglary complaint mentioned in the story had in fact been filed with the police. He also attempted to call appellee on the day the story was to go to press, but, being unable to reach him, went forward with publication, apparently *633with no further attempt at verification . of his facts. The day the story ap- ¡ peared appellee called appellant, and that / conversation led to the inclusion of this item in the column published December 15:
CAREY’S DENIAL — Our report on the circumstances surrounding the reported burglary of a box of “miscellaneous items” from United Mine Workers headquarters has drawn a belated but angry denial from the union’s general counsel Ed Carey. “A contemptible, despicable lie,” said Carey. Our report was based upon information supplied by eyewitnesses, and we will not retract.
Since this second story indicated that the first had been based on eyewitness observation, appellant was asked the identity of those sources. Although appellant refused, as directed by his counsel, to give their names, he did indicate that there was more than one such informant, and that they were UMWA employees.1 Appellant also disclosed in the course of the deposition that he did not know the date on which the activities described by him had occurred, but said it could have been any time during the six weeks immediately prior to publication. The claimed eyewitnesses had provided no written statements or affidavits, nor did appellant know of any writing or recording summarizing their accounts of these incidents. He was unsure whether he had taken notes of the revelations made to him by his informants, and, if so, whether he had preserved them.
Appellee then made a motion under Rule 37(a), Fed.R.Civ.P., to compel appellant to reveal the names of the eyewitnesses. Appellant filed an opposition to this motion, the first ground of which was that the “generic question” raised by the motion was pending before the Supreme Court on petitions for writs of certiorari in the Branzburg trilogy of cases; and that because of “the significant posture of the question at issue before the Supreme Court,” the motion should be denied pending disposition of the writs.2 The other grounds advanced by 'appellant were the constitutional claim of privilege, and the assertion that the information sought was not relevant to appellant’s case or material to his proof.
The Pretrial Examiner recommended that appellant be required to answer the questions concerning the eyewitnesses; and the District Court, disallowing appellant’s objections to the recommendation, entered an order directing him to do so, with a stay provided pending determination of. an appeal. This appeal is from that order.
II
There is no dispute before us as to the standard of proof which appellee must *634meet if he is ultimately to win his ease. That standard is the rigorously demanding one of New York Times Co. v. Sullivan, 376 U.S. 254, 84 S.Ct. 710, 11 L.Ed.2d 686 (1964), where the Supreme Court established the principle that a civil libel plaintiff who is a public figure must show that the statement at issue was published with actual malice or in reckless disregard of the truth.3 In the context of an asserted newsman’s privilege to protect confidential news sources, the Sullivan rule is a source of tension. On the one hand, the Court’s concern that the spectre of potential libel actions might .have an inhibiting effect on the exercise of press freedom militates against compulsory disclosure of sources. Contrarily, the heavy burden of proof imposed upon the plaintiff in such a case will often make discovery of confidential sources critical to any hope of carrying that burden.
Some six years before Sullivan markedly altered the ground rules for recovery in civil defamation actions, there was decided the case of Garland v. Torre, 259 F.2d 545 (2d Cir. 1958), cert. denied, 358 U.S. 910, 79 S.Ct. 237, 3 L.Ed.2d 231 (1958). The opinion for the court was written by Circuit Judge (now Mr. Justice) Stewart, sitting in the Second Circuit by designation. In that case Judy Garland sued Columbia Broadcasting System, Inc., alleging that the latter had made false and defamatory statements about her, and had authorized or induced their publication in newspapers. An example of such publication was attached to the complaint in the form of some paragraphs appearing in a TV-radio column written by Marie Torre and printed in the New York Herald Tribune. The column contained several statements about Garland which Torre in her column attributed to a CBS “network executive.”
In an attempt to learn the name of the alleged detractor, plaintiff deposed three CBS executives without success. In a deposition taken by plaintiff Garland, the columnist refused to disclose her sources and was sentenced to 10 days in jail for criminal contempt. On appeal the Second Circuit upheld the District Court, concluding that resolution of the First Amendment issue in the case before it, although perhaps “delicate,” was not “difficult.” See Schneider v. State of New Jersey, 308 U.S. 147, 161, 60 S.Ct. 146, 84 L.Ed. 155 (1939).
The key factor which the Second Circuit identified as allowing it to move confidently to this conclusion was that the “question asked of [Torre] went to ¡the heart of the plaintiff’s claim.” The ¡decisional process with respect to the ¡¡constitutional issue before it, said the “court, involved a determination of “whether the interest to be served by compelling the testimony of the witness in the present case justifies some impairment of this First Amendment freedom . . . ” Conceding that freedom of the press is “basic to a free society,” the court went on to say that “basic too are courts of justice, armed with the power to discover truth”, and that the “concept that it is the duty of a witness to testify in a court of law has roots fully as deep in our history as does *635the guarantee of a free press.” The court went on to conclude that “[I]f an additional First Amendment liberty— the freedom of the press — is here involved, we do not hesitate to conclude that it too must give place under the Constitution to a paramount public interest in the fair administration of justice.”4
In striking the constitutional balance contemplated in Garland, it could perhaps be argued that, although the Sullivan decision did not eliminate civil libel suits entirely, it has so downgraded their social importance that a plaintiff’s interest in pressing such a claim can rarely, if ever, outweigh a newsman’s interest in protecting his sources. The tenor of the Court’s opinion in Sullivan may be thought to reflect an attitude to- ^ ward libel actions palpably different/) from its approach to grand jury pro-' ceedings in Branzburg. There is, however, the matter of the Court’s continuing gost-Sullivan citations of Garland.5 This strongly suggests the continuing vitality of the latter case, and negates any inference that the Court does not; consider the interest of the defamed| plaintiff an important one.
Branzburg’s lengthy discussion of a newsman’s duty to testify before a grand jury undoubtedly has implications with respect to the deference to be accorded a newsman’s claim of privilege in other areas as well. Although the differences between civil and criminal proceedings distinguish Branzburg from *636the case before us,6 we cannot ignore the fact that the interests asserted by the newsmen in the Branzburg trilogy of cases were not accorded determinative weight by five members of the Court.
Although these cases provide some guidance as to the importance to be assigned to each of the competing interests, they do not seem to disturb the basic balancing approach set forth in Garland. Indeed, the Branzburg result appears to have been controlled by the vote of Justice Powell.7 His concurring opinion states:
The asserted claim to privilege should be judged on its facts by the striking of a proper balance between freedom of the press and the obligation of all citizens to give relevant testimony with respect to criminal conduct. The balance of these vital constitutional and societal interests on a ease-by-ease basis accords with the tried and traditional way of adjudicating such questions. (Footnote omitted). 408 U.S. at 710, 92 S.Ct. at 2671.
Hopefully anticipating a different result in Branzburg, appellant was content to present the case to us upon the theory that the First Amendment left no room, under any circumstances, for compelling a newsman to identify his source. That is clearly not the law after Branzburg with respect to criminal proceedings, and it appears to us that Branzburg, in language if not in holding, left intact, insofar as civil litigation is concerned, the approach taken in Garland. That approach essentially is that the court will look to the facts on a case-by-case basis in the course of weighing the need for the testimony in [ question against the claims of the newsman that the public’s right to know is ' impaired.8
Ill
Turning to the facts of the case before us, the information sought appears to go to the heart of appellee’s libel action, certainly the most important factor in Garland.9 It would be exceedingly *637difficult for appellee to introduce evidence byond his own testimony to prove that he did not, at any time of day or night over an indefinite period of several weeks, remove boxfuls of documents from the UMW offices. Even if he did prove that the statements were false, Sullivan also requires a showing of malice or reckless disregard of the truth. That further step might be achieved by proof that appellant in fact had no relia-, ble sources, that he misrepresented the) reports of his sources, or that reliance! upon those particular sources wasl reckless.10
Knowledge of the identity of the alleged sources would logically be an initial element in the proof of any of such circumstances. Although it might be if possible to submit the question of malice to the jury simply on the basis of the conflicting allegation^ of the parties, that procedure would seem to provide the plaintiff little/prospect of success in view of his heavy burden of proof.11 Consequently, we find that the identity of appellant’s sources is critical to appellee’s claim.
In Garland the court was unable to say that the plaintiff’s claim was frivolous. Neither can we conclude on the basis of the record before us that appel-lee’s claim is without merit. In that respect this case is distinguishable from Cervantes v. Time, Inc., 464 F.2d 986 (8th Cir. 1972), which involved a libel action by the Mayor of St. Louis against Life magazine. After a prolonged and detailed investigation, Life published an article representing that the Mayor had maintained business and social ties with organized crime. Although the article consisted of 87 paragraphs generally containing statements detrimental to the Mayor’s public image, his complaint challenged only four of those paragraphs, while the truth of the rest was either admitted or not expressly denied.
The Mayor sought by deposition to discover confidential sources upon whom Life had relied for part of the story. The defendant publisher refused to disclose that information and countered with a motion for summary judgment. That motion was accompanied by extensive affidavits detailing the prolonged and careful research, and numerous other sources, both public and private, upon which the article was based. The court granted summary judgment for the defendant on the ground that, regardless of the identity of the confidential sources, the plaintiff would be unable to establish malice. It reasoned that the extensive documentation and uncontro-verted accuracy of the bulk of the article, combined with the evidence as to the prolonged, careful, and comprehensive nature of the defendant’s investigation, made it so unlikely that the plaintiff could succeed in his suit that compulsory disclosure of the confidential sources was unwarranted.
There has, of course, been no motion by appellant for summary judgment in the case before us, and thus no showing of facts bearing upon the likelihood that appellee will be able to meet the Sullivan standard. The present pre-trial record, in any event, discloses no wide-ranging and thorough investigatory effort by appellant comparable to that demonstrated in Cervantes. Aside from the confirmation of the burglary report, his description of appellee’s actions appears to have been based solely on the confidential sources in question. In Cervantes the article had been prepared over a period *638of several months by numerous people who had “spent countless hours corroborating and evaluating this data.” 464 F.2d at 994. In contrast to that procedure, after appellant attempted unsuccessfully to reach appellee by one telephone call, he went ahead with publication immediately without waiting even one day to attempt to obtain appellee’s comments. Further, his only action when appellee denied the story was to publish that denial along with a refusal to retract.12
We do not, of course, mean to intimate any opinion as to whether appellant’s investigation was in fact adequate, or whether appellee will ultimately prevail. We say only that the facts disclosed by the record before us at this time are inadequate to support a conclusion that appellee is so unlikely to meet the admittedly heavy Sullivan burden that no purpose would be served by disclosure of the identity of the sources.
. There is, finally, the matter of the possible availability of the information in question from someone other than appellant. The values resident in the protection of the confidential sources of newsmen certainly point towards compelled disclosure from the newsman himself as normally the end, and not the beginning, of the inquiry. In Garland the court took into account the fact the plaintiff had deposed three CBS executives without success before turning to Torre. In Baker, note 9 supra, the district court, in exercising its discretion to deny compelled revelation by the writer, had found that there were other sources that might have disclosed the true name of the pseudonymous informant.
Our information on this score is confined to what appears in appellant’s deposition. There appellant testified that the eyewitnesses were all employees of UMWA. When asked as to their number, this colloquy occurred:
THE WITNESS: There is one that I know of personally and I understand that there were others.
BY MR. FORESTER:
Q. Others being at least two more that we are talking about ?
A. At least one more.
As to the time when the observations occurred, appellant testified that his understanding was that “this was something that had been done over a period of time, more than one occasion;” and he was unable to give any idea of the number of such occasions because his sources had not been specific about that. Appellant stated his understanding to be that the observations occurred in the daytime during normal working hours, but he was unable to say precisely where in the building they took place.
We think it may be assumed that the national offices of the UMWA are manned by a very substantial number of employees. It is also clear from the foregoing that the observations in question could have been made by anyone from an office boy to a top officer, and in any part of the building. We do not think that the concept of exhaustion of remedies relevant here is invoked by guide marks as vague as these. Appel*639lant’s own information as to the circumstances of the observations was so imprecise as to afford appellee no reasonable basis to know where to begin.
These facts are quite unlike those in Garland, where the plaintiff testified in a deposition that she did not know who the CBS “network executive” was who was referred to in Torre’s column, but she had been told that it was “either Lester Gottlieb or Hubbell Robinson.”13 Neither does this record appear to be like that in Baker where there were some sixty defendant real estate agents charged with certain discriminatory practices, and the informant was described as a real estate agent himself who had engaged in the same practices at the same place at the same time; the court obviously saw no reason why the defendants themselves could not be deposed for the same information.
The courts must always be alert to the¡ possibilities of limiting impingements* upon press freedom to the minimum j and one way of doing so is to make comj pelled disclosure by a journalist a last resort after pursuit of other opportunities has failed. But neither must lit-gants be made to carry wide-ranging and onerous discovery burdens where the path is as ill-lighted as that emerging from appellant’s deposition.
IV
It is important to remember that this matter is before us on an appeal from an order of the District Court overruling an objection to that part of the pretrial examiner’s recommendation that appellee’s Rule 37 motion be granted and that appellant, “either on further deposition or in writing, identify the ‘eye-witnesses’ referred to by him . . .” Before discovery began, appellant had filed an answer denying liability. There have been no other proceedings thus far by way of motion for judgment or otherwise, and of course there has been no trial.
What we have decided — and all that we have decided — is that the District Court cannot, on the limited record before us, be said to have abused the discretion vested in it to grant or to deny a motion to compel discovery under Rule 37. We have rejected the only contention made to us by appellant, and that was the pre-Branzburg claim that there either is, or should be, an absolute First Amendment barrier to the compelled disclosure by a newsman of his confidential sources under any circumstances. That was not, in our view, the law before Branzburg, and it is certainly not the law after, in either civil or criminal proceedings.
Affirmed.