Robert M. Lucy, St. Louis, Mo., with whom Daniel C. Schwartz and Penny Q. Seaman, Washington, D.C., were on the brief, for appellant.

Freddi Lipstein, Atty., Dept. of Justice, Washington, D.C., with whom Richard K. Willard, Asst. Atty. Gen., Dept. of Justice, Joseph E. diGenova, U.S. Atty., and Barbara L. Herwig, Atty., Dept. of Justice, Washington, D.C., were on the brief, for appellee.

Before WILKEY*, WALD, and SCALIA, Circuit Judges.

Opinion for the Court filed by Circuit Judge WALD.

WALD, Circuit Judge:

McDonnell Douglas Corporation ("MDC") appeals from an order of the United States Court for the District of Columbia denying MDC's motion to compel discovery and quashing subpoenas duces tecum directed to the United States Departments of Defense ("DOD") and State ("State"). We affirm the district court's order as it applies to the subpoena directed to DOD, but vacate and remand the order as it applies to the subpoena directed to State.

I. BACKGROUND

MDC's pursuit of discovery against DOD and State stems from its ongoing litigation with Northrop Corporation ("Northrop") regarding the development and sale of variations of the YF-17 military aircraft, Northrop Corp. v. McDonnell Douglas Corp., CA No. 79-04145R (C.D.Cal. filed Oct. 26, 1979). The United States is not a party to this litigation. The facts of the underlying action, which is being heard by the United States District Court for the Central District of California, are set out in Northrop Corp. v. McDonnell Douglas Corp., 498 F.Supp. 1112 (C.D.Cal.1980), rev'd in part and remanded, 705 F.2d 1030 (9th Cir.), cert. denied, --- U.S. ----, 104 S.Ct. 156, 78 L.Ed.2d 144 (1983); we present here only a brief summary of the facts relevant to this appeal.

Northrop and MDC, two major defense contractors, entered into a "teaming agreement" to develop two variations of the YF-17, one to be land-based and the other for use on aircraft carriers (the latter was purchased by the U.S. Navy and designated the F-18). According to the terms of the agreement, MDC was to be the prime contractor for the domestic and foreign sales of the aircraft carrier variation, with Northrop acting as a substantial subcontractor, and Northrop was to be the prime contractor for the land-based variation of the YF-17. While MDC has sold its F-18s to the U.S. Navy and at least three foreign governments, Northrop has not been able to sell any of its variations of the aircraft. Brief for Appellant at 5-6. Northrop sued MDC claiming, inter alia, that MDC violated the terms of the teaming agreement, interfered with Northrop's efforts to sell its version of the YF-17, and violated the antitrust laws. MDC has asserted as part of its defense that Northrop's inability to sell any of the land-based YF-17s is the result of actions by the United States government, not MDC. In furtherance of that defense, on December 21, 1983, MDC subpoenaed DOD, State, and the Departments of the Air Force and the Navy for documents relating to sale of various military equipment to Iran, Canada, Australia, Spain, Turkey, Sweden, Israel and the Federal Republic of Germany, generally covering the period from January, 1977, to the present.1

The Departments of the Air Force and the Navy complied with the subpoenas to MDC's satisfaction. DOD produced 3000 pages of documents to MDC and, on March 26, 1984, claimed that 1200 more pages of responsive documents were privileged as state or military secrets. Brief for Appellant at 14. Initially, State also produced some documents, but then on January 31, 1984, filed an objection to the subpoena on the grounds of burdensomeness. State ultimately claimed that 967 cubic feet of documents would have to be searched to comply with the subpoena, and that such a search would involve hundreds of worker hours. Further, State claimed that many of the responsive documents would be classified and subject to the state secrets or deliberative process privileges, and that a declassification review would involve yet additional hundreds of worker hours. Appendix to Brief for Appellant ("Appendix") at 272, 277.

In response to DOD's claim of privilege and State's claim of oppressiveness, MDC moved to compel production of the documents from both parties. The district court held a one-day hearing on the motion. In a one-page order the court denied MDC's motion to compel and quashed the subpoenas. This appeal followed.

II. ANALYSIS

MDC claims that the district court erred in quashing the subpoenas directed to DOD and State.2 As to DOD, MDC contends that the district court, given MDC's representations of the importance of the documents to MDC's defense, should have conducted an in camera review of the documents to determine whether the state secrets privilege was properly invoked.3 MDC further asserts that the district court should not have permitted DOD's assertion of the state secrets privilege because it did not establish the requisite likelihood that harm would result in the event of disclosure.4 As to State, MDC appeals the court's order quashing the subpoena on grounds of burdensomeness. According to MDC, State did not adequately demonstrate the oppressiveness required to quash the subpoena, particularly in light of MDC's asserted need for the documents, the unavailability from other sources of the information contained in those documents, and the complexity of the underlying Northrop litigation.

In evaluating a trial court's exercise of discretion in discovery matters, we have observed that

[a] district court has broad discretion in its resolution of discovery problems that arise in cases pending before it.... [T]he scope of appellate review is equally narrow when the discovery pertains to litigation pending elsewhere.

In Re Multi-Piece Rim Products Liability Litigation, 653 F.2d 671, 679 (D.C.Cir.1981). We may reverse the trial court only if it has abused its discretion; that is, if its actions were clearly unreasonable, arbitrary or fanciful. Id. Giving the district court due deference, we find it properly exercised its discretion in accepting DOD's claim of privilege, but that its action in quashing the State subpoena in its entirety, without adequate consideration of whether it might be modified so as to diminish the burden, was unreasonable and an abuse of discretion.5

A. DOD's Claim of Privilege

The "state secrets" privilege asserted by DOD is a privilege developed in common law protecting information vital to the nation's security or diplomatic relations.6 See Advisory Comm. Note to Proposed Fed.R.Evid. 509, reprinted in 56 F.R.D. 194, 252 (1972); E. Cleary, McCormick on Evidence Sec. 107 (3d ed. 1984). It is an absolute privilege which, when properly asserted, cannot be compromised by any showing of need on the part of the party seeking the information. The seminal judicial statement on the privilege appears in United States v. Reynolds, 345 U.S. 1, 73 S.Ct. 528, 97 L.Ed. 727 (1953). Reynolds establishes the procedure which must be followed for the privilege to be properly invoked: " There must be a formal claim of privilege, lodged by the head of the department which has control over the matter, after actual personal consideration by that officer." Id. at 7-8, 73 S.Ct. at 531-532 (footnotes omitted). The inherent dilemma presented by the privilege, as recognized by the Reynolds Court, is that a court must "determine whether the circumstances are appropriate for the claim of privilege, and yet do so without forcing a disclosure of the very thing the privilege is designed to protect." Id. at 8, 73 S.Ct. at 532 (footnotes omitted).

We reiterate at the outset that a party's need for the information is not a factor in considering whether the privilege will apply. As we have stated,

courts in evaluating claims for the privilege may take cognizance of the need for the information demonstrated by the party seeking disclosure, [but] such need is a factor only in determining the extent of the court's inquiry into the appropriateness of the claim. Once the court is satisfied that the information poses a reasonable danger to secrets of state, "even the most compelling necessity cannot overcome the claim of privilege."

Halkin v. Helms, 690 F.2d 977, 990 (D.C.Cir.1982) (footnote omitted) ("Halkin II ") (quoting Reynolds, 345 U.S. at 11, 73 S.Ct. at 533). Therefore, MDC's representations of the importance of the government documents to its defense in the Northrop litigation are relevant only to the question of whether the district court adequately assessed DOD's assertion of the privilege. Reviewing the documents submitted to the court in support of DOD's claim, we find that the district court did not abuse its discretion in holding that the privilege was adequately asserted.

According to the procedure established by Reynolds, DOD properly made its claim to the state secrets privilege. In an affidavit, the head of the department, Secretary of Defense Weinberger, stated that he had reviewed a representative sample of the documents as well as affidavits of staff members who had received all of the documents, and based on this knowledge he "assert[ed] a formal claim of privilege in order to protect certain military and state secrets relating to the national defense and the national security of the United States ...." Affidavit and Claim of Privilege of the Secretary of Defense ("Weinberger Affidavit") p 2, Appendix at 282-83. The affidavit states that "[t]he Department of Defense has in its possession seven linear inches of classified documents responsive to the subpoena." Weinberger Affidavit p 5, Appendix at 284. The documents, "all of which have been classified pursuant to appropriate executive orders and DOD regulations," fall into the general categories of: Communications with foreign government officials (letters and memoranda of conversations); letters between DOD and the Secretary of State or the White House; studies of force structures of foreign countries; DOD briefing books and papers for meetings with foreign government officials; internal papers and recommendations relating to military aircraft sales to foreign countries, sales terms, coproduction arrangements, and related matters; Defense Intelligence Agency reports; and correspondence with Members of Congress regarding the sale of military aircraft. Weinberger Affidavit p 6, Appendix at 285. Secretary Weinberger asserts that "disclosure of this material to McDonnell Douglas and Northrop would have specific adverse and irreparable effects upon the national security and international relations of the United States." Weinberger Affidavit p 7, Appendix at 286. The harms alleged in the affidavit that could result from disclosure are: The impairment of diplomatic relations if the confidence of communications with foreign governments is breached; adverse effects on our relations with Iran; adverse effects on relations with countries getting less favorable treatment on sales terms for military equipment; revelation of military secrets and defenses of foreign countries which in turn threatens the security of the United States; and exposure of sources of intelligence information. Each of the alleged harms is correlated to the document over which the privilege is claimed. Weinberger Affidavit paragraphs 7-10, Appendix at 286-89.

Relying on Reynolds, MDC contends that the district court did not examine the privilege claim carefully enough, given the importance of the documents to MDC's defense. See Reynolds, 345 U.S. at 11, 73 S.Ct. at 533 ("In each case, the showing of necessity which is made will determine how far the court should probe in satisfying itself that the occasion for invoking the claim is appropriate."). The apparent basis for MDC's contention is that the district court denied the motion to compel on the day immediately following its one-day hearing. Certainly MDC's claimed need for the documents is substantial; asserting a defense of government action will be more difficult without the government's own records of that action. But even if the lack of these documents made MDC's case impossible, MDC would still not be entitled to overcome a valid assertion of the state secrets privilege. See Halkin II, 690 F.2d at 997-98, 999-1000.7 And while the order came swiftly, that in itself is no cause for presuming it was not the product of a considered judgment.8

MDC suggests that before ruling on the motion to compel, the district court should have conducted an in camera review of the documents over which privilege was claimed. While we have recognized the value of such examinations in the context of privilege claims, see, e.g., Halkin v. Helms, 598 F.2d 1, 5 (D.C.Cir.1978) ("Halkin I "), the procedure is not required as a matter of course in a claim of the state secrets privilege. Reynolds explicitly addressed this issue.

It may be possible to satisfy the court, from all the circumstances of the case, that there is a reasonable danger that compulsion of the evidence will expose military matters which, in the interest of national security, should not be divulged. When this is the case, the occasion for the privilege is appropriate, and the court should not jeopardize the security which the privilege is meant to protect by insisting upon an examination of the evidence, even by the judge alone, in chambers.

345 U.S. at 10, 73 S.Ct. at 533.

Our opinion in Ellsberg v. Mitchell, 709 F.2d 51 (D.C.Cir.1983), cert. denied, --- U.S. ----, 104 S.Ct. 1316, 79 L.Ed.2d 712 (1984), established a sliding scale to determine when a court may, or should, make an in camera examination of material over which a claim of the state secrets privilege is made:

When a litigant must lose if the claim [of privilege] is upheld and the government's assertions are dubious in view of the nature of the information requested and the circumstances surrounding the case, careful in camera examination of the material is not only appropriate ... but obligatory.

....

When the litigant requesting the information has made only a trivial showing of need for it and the circumstances of the case point to a significant risk of serious harm if the information is disclosed, the trial judge should evaluate (and uphold) the privilege claim solely on the basis of the government's public representations, without an in camera examination of the documents.

Id. at 59 nn. 37-38 (citations omitted).

MDC's claim falls somewhere in between the two extremes. The government's assertions are more than dubious, and MDC's need for the information is more than trivial, though not absolute. Although an in camera inspection might have been appropriate, we cannot say that the trial court erred when it decided the balance tipped in favor of the government, and quashed the subpoena without conducting an in camera review of the documents.

The Secretary's affidavit alone, MDC contends, does not adequately demonstrate that the national security or international relations will be harmed by the disclosure of the information MDC seeks. MDC suggests that even though the information involved is sensitive, there is no threat that its production in this litigation will do harm because the disclosure will be limited to participants with adequate security clearances. The trustworthiness of the litigants, however, is not always dispositive in cases such as this.

There is no single rule which will determine when disclosure will present a harm to national interests. The grounds for finding such harm will vary from case to case, depending upon the nature of the information at issue, to whom it will be disclosed, and the means available to limit its dissemination. The latter two circumstances are not decisive in the present case. The trial court in the underlying action has issued a protective order establishing procedures for handling information related to the case which may not be publicly disclosed. Northrop and MDC have subscribed to the terms of that order. See Appendix at 203-08. To the extent that sensitive information can be protected in litigation, then, we can assume it would be protected here. The party seeking the information, MDC, and its adversary, Northrop, have in the past been entrusted by the government with classified information necessary to their performances of government contracts. Whether this past practice means they should also be entrusted with classified policy information related to their dispute is a matter of contention between MDC and the government which we do not find it necessary to resolve, because we believe that the nature of the information at issue here justified the district court's decision to accept DOD's claim of privilege.

The documents MDC seeks concern foreign governments' plans and policies for their national defense systems, and our government's involvement in those areas. DOD credibly asserts that such inter-government communications about defense systems are made with every expectation of confidentiality on both sides. Equally credibly, DOD asserts that our government's ability to conduct relations with these countries in the future would be harmed by the disclosure, however protected, of the most confidential information a country can possess: the details of its national security system. Regardless of the availability of protective orders or "need-to-know" mechanisms, we believe that the district court acted within reason when it decided that this disclosure would present a danger of harm to foreign relations and national security. Without setting forth any absolute rule that communications with or about foreign governments are always immune, we accept as a reasonable justification of the district court's action that in this instance disclosure, regardless of the attendant safeguards, in itself might harm U.S. foreign relations and national security.

According to the standards established by Reynolds and this court's interpretation of Reynolds, all DOD must show is a reasonable danger that harm will result from disclosure. See Ellsberg, 709 F.2d at 58; Halkin II, 690 F.2d at 990; Halkin I, 598 F.2d at 9. It is not necessary for the government to show that harm will inevitably result from disclosure, nor, as MDC argues, is it an essential element that the disclosure be public. "The crucial aspect of [the various] formulation[s] of the test [determining whether the requisite degree of certainty that harm is threatened] is the [court's] willingness to credit relatively speculative projections of adverse consequences." Ellsberg, 709 F.2d at 58 n. 35. And in evaluating these "speculative projections," "[c]ourts should accord the 'utmost deference' to executive assertions of privilege upon grounds of military or diplomatic secrets." Halkin I, 598 F.2d at 9 (quoting United States v. Nixon, 418 U.S. 683, 710, 94 S.Ct. 3090, 41 L.Ed.2d 1039 (1974)).

Deference to the executive's allegations of harm does not mean unquestioning acceptance of every claim of privilege. See Reynolds, 345 U.S. at 9-10, 73 S.Ct. at 532-533 ("Judicial control over the evidence in a case cannot be abdicated to the caprice of executive officers.") We have recognized that drawing the line between an adequate description of the privileged material and divulging the secrets themselves is not a simple task. In Halkin I, we said that simply identifying certain individuals as having been subject to government communications monitoring presented "a reasonable danger that state secrets would be revealed." Halkin I, 598 F.2d at 9. In the present case, identifying documents as being withheld because Country X will discover that it received less favorable terms than Country Y on an arms sale agreement obviously does as much damage as revealing the documents themselves. In sum, DOD has adequately asserted the state secrets privilege, and the district court did not act unreasonably in accepting DOD's claim.9B. State's Claim of Burdensomeness

The district court's order also denied MDC's motion to compel discovery and quashed the subpoena directed against State. We hold that the district court abused its discretion in issuing that order.

A trial court may, on a motion by the party seeking relief, quash or modify a subpoena for the production of evidence, on the ground that the request is unreasonable or oppressive. See Fed.R.Civ.P. 45(b)(1). "What constitutes unreasonableness or oppression is, of course, a matter to be decided in the light of all the circumstances of the case ...." 5A Moore's Federal Practice p 40.05 n. 44. A trial court has broad, but not unlimited, discretion in evaluating the circumstances of a case when considering quashing a subpoena on grounds of oppressiveness. It must carefully examine the circumstances presented to it and, when appropriate, consider the possibility of modifying the subpoena rather than quashing. The burden of proving that a subpoena is oppressive is on the party moving to quash. Westinghouse Electric Corp. v. City of Burlington, Vt., 351 F.2d 762 (D.C.Cir.1965). This is a heavy burden, and one which State has not met in the present case. Because State's claim of oppressiveness is in large part based on its assertion that many of the documents sought by the subpoena will be privileged, and therefore not subject to compelled production, it must adequately demonstrate that applicable privileges would in fact protect these documents. In addition, the district court must give appropriate consideration to the possibility of modifying the subpoena, to accommodate the interests of both MDC and the government.

We stressed the importance of the modification alternative in our decision in Westinghouse. Westinghouse, a defendant in a treble damages antitrust action, served a subpoena duces tecum on a representative of the U.S. Attorney General, seeking records of complaints from utility owners about antitrust violations by electrical equipment manufacturers. Westinghouse sought this information to dispute plaintiffs' claim that the statute of limitations should be tolled because Westinghouse had fraudulently concealed the alleged violations. The United States was not itself a party to the treble damages action. The government moved to quash the subpoena on the grounds that it was oppressive and that in any case the documents sought were protected by the informer's privilege. The trial court granted the motion to quash, but this court reversed, finding that the record did not support the government's claim of oppressiveness. See id. at 767.

We held in Westinghouse that the government, as the party seeking to quash, had not met its burden of establishing that the subpoena was unreasonably oppressive. The district court had erred because

[t]he two affidavits submitted by [the government] are not sufficient to bear this burden. Appellants [Westinghouse] have expressed their willingness to accept some reasonable effort by the Government, that would be less than a page-by-page search. The lower court, in these circumstances, should have sought some way to accommodate the interests of the defendants herein with the practical problems of searching the Government's voluminous files. Nothing in the record suggests that the possibility of making a less than complete search was explored.

Id. at 766 (footnotes omitted). The Westinghouse court went on to suggest that, as an aid to the trial court in evaluating the oppressiveness of the subpoena, the government could conduct a partial search of its files "to determine how productive and how onerous a search of the complete file would be." Id. at 767.

Westinghouse is relevant to the present case in several ways. First, MDC has, as did appellants in Westinghouse, expressed its willingness to modify the subpoena. See Appendix at 128, 130. The district court was also informed that the California court hearing the underlying action had established an accelerated discovery schedule and was not disposed to grant any further postponements. See Transcript at 7; see also Appendix at 95. This additional factor presented an even stronger case for modification than the circumstances of Westinghouse. To facilitate the progress of the underlying litigation, the district court should have at least explored the possibility of modifying the subpoena in a manner which would meet both MDC's and State's immediate needs.10

Second, Westinghouse made clear that "[t]he burden of proving that a subpoena duces tecum is oppressive is on the party moving for relief on this ground.... The burden is particularly heavy to support a 'motion to quash as contrasted to some more limited protection.' " Westinghouse, 351 F.2d at 766 (quoting Horizons Titanium Corp. v. Norton Co., 290 F.2d 421, 425 (1st Cir.1961)). State has not adequately carried that burden here. Certainly the volume of documents it claims it must search is extraordinary, but volume alone is not determinative. See Democratic Nat'l Comm. v. McCord, 356 F.Supp. 1394, 1396 (D.D.C.1973); 5A Moore's Federal Practice p 45.05 n. 44. State concedes that its files are organized topically and geographically, see Appendix at 263, so there apparently exists some systematic way of conducting the search. According to State, though, the compelling reason for finding the subpoena oppressive is that "it would be necessary to claim privilege over a high proportion of the documents" once retrieved. Brief for Appellees at 43. To support this reasoning, State must demonstrate, with more certainty than it has here, that privileges would in fact be claimed over a substantial percentage of the documents.

The privileges which State claims will protect many of the documents produced, the state secrets and deliberative process privileges, are narrowly drawn privileges which must be asserted according to clearly defined procedures. See supra p. 400. The procedure for claiming the state secrets privilege includes a statement by the head of the department responsible for the documents that they have been examined and that disclosure would be against the public interest. See Reynolds, 345 U.S. at 7-8, 73 S.Ct. at 531-532; Ellsberg, 709 F.2d at 56-57. Clearly, documents which have not yet been retrieved have not yet been examined or considered by the head of the department. It is premature for State to assert or even insinuate a claim of the state secrets privilege over these documents.

It is equally untenable to claim the deliberative process privilege over the documents. That privilege, unlike the absolute state secrets privilege, is relative to the need demonstrated for the information. See Carl Zeiss Stiftung v. V.E.B. Carl Zeiss, Jena, 40 F.R.D. 318, 327 (D.D.C. 1966), aff'd, 384 F.2d 979 (D.C.Cir.), cert. denied, 389 U.S. 952 , 88 S.Ct. 334, 19 L.Ed.2d 361 (1967). The litigant's need for the information cannot be balanced against its sensitive and critical role in the government's decision making process without any indication of what that information is. Any claim that the documents would be protected by this privilege is purely speculative.11

Third, in Westinghouse and Freeman v. Seligson, 405 F.2d 1326 (D.C.Cir.1968), we suggested the possibility that conducting partial searches, or samplings, of voluminous files containing potentially privileged information would be an appropriate aid to evaluating oppressiveness claims. This proposal is particularly well-suited to the difficult problem posed by this case: How can a court evaluate the oppressiveness of a subpoena when the requesting party has demonstrated the relevance of the documents and a need for them, and the objecting party asserts that production of the document will be burdensome and not fruitful for the requesting party because privilege will make many nondiscoverable? To force State to conduct a complete search and then raise a formal claim of privilege would, in this instance, defeat the purpose of allowing oppressiveness as a defense to subpoenas issued under Fed.R.Civ.P. 45. A sampling procedure appears to us to be a sound alternative to the illogical alternative of requiring State to conduct a full search of its files. Although State presented the district court with the results of a sample search of its files, see Brief for Appellees at 42; Appendix at 271, we find that this sample fails in at least four ways.

First, State has not indicated how the files examined are representative of the files identified in the subpoena. The subpoena requests documents from four offices at State;12 yet the declaration of the person reviewing the sample for classification purposes does not indicate which office(s) the sample was drawn from, only that it consisted of "several hundred readily available responsive documents." Declaration of Thomas W. Ainsworth, Appendix at 276. The sample may in fact represent a cross-section of all the named offices, or it may represent only one. To adequately evaluate the sample, a court must know that it is representative of all the sources named in the subpoena. Second, no precise indication is made of what percentage of the documents in the sample are likely to be privileged. The declaration offered by State only tells us that "the sample contains a high proportion of currently classified national security information...." Appendix at 276. To be of any value to a court assessing a sampling, such a statement would have to indicate what percentage of the documents would be subject to the privilege, recognizing that not all classified information will necessarily be withheld from production.

Third, State does not indicate why the sample can be expected to contain a typical proportion of documents likely to be privileged. It seems entirely plausible that some files will consist almost exclusively of documents containing state or military secrets, while other files will be almost entirely devoid of such materials. Selecting either type of file as the exclusive basis for a sample would distort the picture of the overall fruitfulness of a search. Because this variable can exist within a single office as well as in the department as a whole, State must indicate that the sample it offers contains a typical proportion of potentially privileged documents based on the universe it represents.

In the circumstances of the present case, a valid sample on which a trial court could make an adequate assessment of the burden of full compliance with a subpoena would have to include at least the following information. First, it must adequately describe and demonstrate how the sample is representative of all the sources identified in the subpoena. If a variety of sources are named, as in MDC's subpoena to State, this may require independent samplings for each source to properly represent the underlying universe of documents. Second, the sample must state what percentage of documents are likely to be privileged. The court reviewing the sample can then determine if that proportion is "high" enough to support a claim of burden. Third, the sample must demonstrate that the percentage of potentially privileged documents in the sample can be expected to represent the percentage in the underlying universe. Again, this may mean that a single sample will not adequately represent the composition of all the files subpoenaed. If one source has a particularly high (or low) concentration of potentially privileged documents, it might need to be considered separately from any "overall" claims. Additionally, a valid sample should include estimates of the time and labor involved in complying with the subpoena and the basis of these estimates. In the sample offered to the district court, State did estimate the amount of labor that retrieval and review of the documents requested by MDC would take. See Appendix at 272-73, 277.

Establishing the representativeness of the sample does not end State's burden. State's present projection of the state secrets privilege is made by individuals not qualified to formally assert the state secrets privilege. Their speculation that the Secretary would assert a claim of privilege if presented with the documents is not acceptable; this fourth defect alone defeats any validity an otherwise proper sample might have.

The direct involvement of the head of the department is essential to a valid sample. In this instance the Secretary would have to review the possibly privileged documents in the sample and formally assert that he would make a claim of privilege over them if they were to be produced. As with an actual assertion of privilege, the Secretary's statement would have to include a description of the documents he has considered and the potential threat posed by the disclosure of the information. Additionally, there should be a statement by an appropriate official describing the basis for believing that similarly privileged documents exist in the files not searched.

Clearly, this sampling procedure does not replace the requirements established in Reynolds for a formal claim of the state secrets privilege. To make a proper claim of the military or state secrets privilege, State would have to make a complete examination of its files, and present the court with a formal claim by the Secretary that after considering the documents he has determined that the privilege should be invoked.13 A proper claim of privilege, if accepted by the court, would absolutely protect the documents from discovery, regardless of the degree of MDC's need for them. See Reynolds, 345 U.S. at 11, 73 S.Ct. at 531. The sampling procedure we set forth is a mechanism by which a court can evaluate a claim of oppressiveness based on both the large number of files to be searched and the likelihood that such a search will not be productive because so many of the materials sought will be privileged. Unlike the state secrets privilege, a claim of oppressiveness is not an absolute protection against discovery. The need of the party seeking the documents is a relevant factor in considering a claim of oppressiveness, see, e.g., Multi-Piece Liability Litigation, 653 F.2d at 679-80, and a case may arise where the need is great enough to overcome a claim such as State's here. In that instance it would be necessary for the executive department involved, to protect any state or military secrets contained in its files, to follow the Reynolds procedure and make a formal claim of privilege. We do not at this point express any opinion on whether MDC has established such a need.

As in Westinghouse, "[w]e do not now hold that the broad subpoena sought by the defendants should be enforced; but we instruct the trial judge to reconsider ... the possibility of reaching an accommodation between [the parties in the discovery action]." Westinghouse, 351 F.2d at 765. At least two options are available to the district court in its reconsideration of State's claim of oppressiveness. First, it may order a modification of the subpoena, based on the facts and claims the court has already heard. There seem to be ample grounds for immediately effecting some form of modification. See supra p. 18 & n. 10. Second, the district court may allow State to clarify the basis of its existing sample and submit a statement by the Secretary affirming his intent to assert privilege as described earlier, or give State leave to conduct a new sampling according to the guidelines set out in this opinion. On the basis of either action by State, the district court may decide to modify or quash the subpoena, or compel its enforcement.

As the Northrop-MDC action illustrates, discovery has become a major battleground in litigation. Massive discovery requests are unfortunately now part of the standard tactical maneuvering of parties immersed in the adversarial process. While we fully recognize the burden of imposing on a non-party the effort and expense of discovery, particularly when the expense will be borne by the taxpayers, we also recognize that "the paramount interest of the Government in having justice done between litigants in Federal courts militates in favor of requiring a great effort on its part to produce any documents relevant to a fair termination of th[e] litigation." Westinghouse, 351 F.2d at 767.

Accordingly, the inconvenience of MDC's discovery request on State is not reason enough, alone, to quash the subpoena. See id. A reasonable inconvenience must be borne to further the goals of discovery--the making available to litigants all relevant and available information. See Fed.R.Civ.P. 26, advisory committee notes on 1983 amendments. Whether searching 967 cubic feet of documents is reasonable must be determined according to the facts of the case. As we stated in Westinghouse, "[t]he fact that these are very important cases with large sums of money at stake is relevant in determining the reasonableness of the subpoena." Westinghouse, 351 F.2d at 767. The district court should carefully examine both MDC's need for discovery and the extent of the burden on State when it reconsiders the motion to compel, paying special attention to the possibility of modifying the subpoena to accommodate the needs and interests of both MDC and State.III. CONCLUSION

The order of the district court quashing the subpoenas is affirmed as it applies to the DOD subpoena. The order is vacated as it applies to the State subpoena, and remanded for further proceedings.

So ordered.

* Circuit Judge Wilkey considered this case prior to his taking Senior status

1 The precise dates varied between countries. Documents relating to Iran were requested only for the period of January 1974 to December 1978

2 As a preliminary matter, we address a jurisdictional question raised at oral argument. The court, sua sponte, inquired of counsel whether sovereign immunity might bar an action to enforce a subpoena directed against the government when, as here, the government is not itself a party to the litigation. Supplemental briefs on this issue were submitted, but neither side took the part of sovereign immunity. As far as the briefs and our independent research disclose, the issue has never been explicitly discussed in the opinion of any federal court. Since at least 1965, however, this court has assumed the nonapplicability of sovereign immunity to such a subpoena. See Westinghouse Electric Corp. v. City of Burlington, Vt., 351 F.2d 762, 767 (D.C.Cir.1965); Freeman v. Seligson, 405 F.2d 1326, 1338 n. 61 (D.C.Cir.1968); Friedman v. Bache Halsey Stuart Shields, Inc., 738 F.2d 1336 (D.C.Cir.1984); Carl Zeiss Stiftung v. V.E.B. Carl Zeiss, Jena, 40 F.R.D. 318 (D.D.C.1966), aff'd mem. 384 F.2d 979 (D.C.Cir.), cert. denied, 389 U.S. 952 , 88 S.Ct. 334, 19 L.Ed.2d 361 (1967). At least the 10th Circuit has done likewise. See Hefley v. Textron, Inc., 713 F.2d 1487 (10th Cir.1983); United States Steel v. Mattingly, 663 F.2d 68 (10th Cir.1980), cert. denied, 450 U.S. 980 , 101 S.Ct. 1515, 67 L.Ed.2d 815 (1981); Saunders v. Great Western Sugar Co., 396 F.2d 794 (10th Cir.1968). Rather, courts have proceeded on the basis that the government has a set of special privileges--e.g., executive privilege, state secrets, deliberative process--which it may invoke to prevent disclosures that would be inimical to national security or its internal deliberations. Commentators have noted that while executive privilege against discovery can be viewed as a modern derivative of sovereign immunity, see Berger and Krash, Government Immunity from Discovery, 59 Yale L.J. 1451, 1459 n. 46 (1950), the two doctrines are distinct in their scope and applicability. See Comment, Discovery of Government Documents Under the Federal Rules, 18 U.Chi.L.Rev. 122, 123 (1950)

We think this not the occasion to enter into a detailed survey of this issue, uninformed by that adversary presentation that usually assists our deliberations. Sovereign immunity, which can be avoided by a device no more complex than pronouncing the suit to be one against the officer rather than his employer, see, e.g., Houston v. Ormes, 252 U.S. 469, 472-73, 40 S.Ct. 369, 369-70, 64 L.Ed. 667 (1920), is a doctrine "rooted in history," Gore v. United States, 357 U.S. 386, 392, 78 S.Ct. 1280, 1284, 2 L.Ed.2d 1405 (1958), whose application depends much more upon tradition than logical analysis. See Scalia, Sovereign Immunity and Nonstatutory Review of Federal Administrative Action: Some Conclusions from the Public-Lands Cases, 68 Mich.L.Rev. 867, 909-20 (1970). In light of the above-cited history of this issue, we find no cause in the present case to upset a steady course of precedent by attempting to graft onto discovery law a broad doctrine of sovereign immunity.

3 MDC did not request an in camera inspection in its motion to compel. Appendix at 90. In camera inspection was requested at the March 29, 1984, hearing on the motion. Transcript of Hearing 2-29-84, at 26 ("Transcript"). At the hearing, MDC also asked the court for leave to depose DOD officials regarding the non-production of certain unclassified documents that MDC believes exist in DOD's files. Id. The court's order does not respond to the latter request; we do not interpret the order as barring MDC from seeking a subpoena to depose the DOD officials as non-party witnesses. See generally, 5A J. Moore, W. Taggart & J. Wicker, Moore's Federal Practice p 45.05 (2d ed. 1984) ("Moore's Federal Practice"); C. Wright & A. Miller, Federal Practice and Procedure Sec. 2458 (1971) ("Wright & Miller")

4 MDC also contends that the "deliberative process" privilege claimed by DOD was not properly invoked, and that even if it were it would not prevail over MDC's need for the documents. Brief for Appellant at 19; see Wright & Miller Sec. 2019. Because we find that the state secrets privilege adequately supports the district court's order, we do not reach the question presented involving the deliberative process privilege

5 We note that the district court's order is an appealable final judgment. An order denying discovery in one district court's jurisdiction arising from litigation pending in another jurisdiction may be appealed without awaiting final judgment on the underlying litigation. Multi-Piece Liability Litigation, 653 F.2d at 676; Westinghouse, 351 F.2d at 765; see also 4 Moore's Federal Practice p 26.83

6 The Federal Rules of Evidence specifically provide that common law governs which privileges are recognized in federal courts. Fed.R.Evid. 501

7 Quashing the DOD subpoena does not make MDC's defense impossible. It has obtained substantial discovery from DOD already, as well as from the Departments of the Air Force and the Navy. It has deposed several DOD officials and is free to continue such depositions. When considering a claim of the state secrets privilege, the availability of alternate means of discovery mitigates a party's level of need for the discovery. Reynolds, 345 U.S. at 11, 73 S.Ct. at 533

8 We note, however, that in litigation as complex and extensive as this, especially when the parties have tended to appeal rulings at every step along the way, it would be helpful on review to have a statement of the district court's reasoning supporting its decision

9 It is irrelevant that the Air Force and the Navy, in response to MDC's subpoenas, have produced classified documents covering matters related to those over which DOD claims the state secrets privilege. It is quite possible that, though related, the nature of the precise information contained in the revealed documents is different from that contained in the privileged documents and justifies different treatment. See Brief for Appellees at 31-32. Moreover, prior disclosure of similar information does not preclude the potential for harm resulting from the present, requested disclosure. In Halkin I, we stated that "[t]he government is not estopped from concluding in one case that disclosure is permissible while in another case it is not." Halkin I, 598 F.2d at 9. Similarly, one department's decision to disclose documents it possesses on a given topic should not compel another department's disclosure when it has decided against such action. In the present instance, it is of interest that the Departments of the Air Force and the Navy produced a large number of documents to MDC, because these actions might indicate that a claim of privilege over similar documents should be examined closely. We are satisfied, though, that DOD has adequately asserted its own privilege claim over documents under its control

10 The California court, on February 28, 1984, ruled that Northrop's claims relating to MDC F-18 sales to Canada, Spain and Australia were precluded by settlement agreements between MDC and Northrop. Northrop's motion to reconsider that ruling was denied on April 16, 1984. Brief for Appellants at 34 n. 8. Apparently, this possible means of modifying the subpoena by eliminating documents relating to those countries was not known by the District Court for the District of Columbia at the March 29, 1984, hearing on the motion to compel. This is clearly a relevant factor for the court to include in its reconsideration of MDC's motion

11 Assertion of the deliberative process privilege, like the state secrets privilege, requires a formal claim of privilege by the head of the department with control over the information. That formal claim must include a description of the documents involved, a statement by the department head that she has reviewed the documents involved, and an assessment of the consequences of disclosure of the information. Garber v. United States, 73 F.R.D. 364 (D.D.C.1976), aff'd on other grounds, 578 F.2d 414 (D.C.Cir.1978). See also Carl Zeiss, 40 F.R.D. 318; Wright & Miller Sec. 2019. As with the state secrets privilege, State's claim of this privilege also fails on this procedural requirement

12 The subpoena is limited to documents

in the possession, custody or control of only the offices located in the greater Washington, D.C. area, as described below...

Department of State

Office of Secretary of State

Office of Deputy Secretary of State

Office of Under Secretary of State for Security Assistance, Science & Technology

Office of Director of Politico-Military Affairs including:

(International Security Policy)

(Munitions Control)

(Security Assistance and Sales)

Appendix at 6.

13 Actual consideration of documents over which the state secrets privilege is claimed is an essential element of the Reynolds procedure which has not been altered by this court's recent decision in Molerio v. FBI, 749 F.2d 815 (D.C.Cir.1984). Molerio held that when the reason for the FBI's decision not to hire an applicant was itself a state secret, all documents that would disclose the reason for that decision are not subject to discovery and need not be actually examined. To the extent that the documents discuss the state secret (the reason for the non-hire decision), they would be privileged; to the extent that they do not discuss the reason for the non-hire decision (which was the subject of the suit), they are irrelevant and not subject to discovery. When a claim is made that documents relevant to an issue, which is not itself privilege, contain privileged information--as State is prospectively claiming here--actual consideration of those documents is still required for a valid assertion of the state secrets privilege