This case concerns a Freedom of Information Act request by the American Friends Service Committee (AFSC) addressed to the Department of Defense (DoD) pursuant to 5 U.S.C. Sec. 552 (1982). DoD denied the request, invoking two exemptions in support of its denial. The first, contained in 5 U.S.C. Sec. 552(b)(1), is the national security exemption. The second, contained in id. Sec. 552(b)(3), concerns material whose withholding has been authorized by another statute. In this case the other statute is 10 U.S.C.A. Sec. 130(a) (West Supp.1987), which authorizes the Secretary of Defense to withhold

technical data with military or space application ... if such data may not be exported lawfully outside the United States without an approval, authorization, or license under the ... Arms Export Control Act.

AFSC brought suit to challenge the DoD's denial of its request. On the strength of three affidavits submitted by DoD, and over a single affidavit submitted by an expert on behalf of the AFSC, the district court granted summary judgment in favor of DoD. AFSC appeals.

We agree with DoD on a number of important points--the validity of its "compilation" theory, the inapplicability to this case of the doctrine of segregability, and the standard of review, which requires that considerable deference be given to DoD's judgments regarding national security. We nonetheless vacate and remand so that the district court can resolve factual disputes in the record. We begin by explaining the nature of the documents sought and withheld. We then turn to the merits of the bases for withholding.

AFSC seeks the disclosure of a series of documents called Technical Abstract Bulletins, or TABs. TABs, which are published biweekly, contain the titles of research reports prepared for DoD by DoD scientists, engineers and contractors. The reports themselves are released to the public unless they must be classified. The first section of the TABs contain

citations to every classified and/or limited distribution technical report by accession number, corporate author, subject category designators, ... title descriptive note, originator/sponsor, report numbers, author, date, number of pages, project/task numbers, contract number, report classification, distribution list, descriptors of the report's contents, and an abstract or summary of the report's contents. The second section of each TAB consists of eight indexes to the technical reports described in the first section, including indexes by subject category ..., subject and author.

Appellee's Br. at 5. The TABs do not contain entries for reports classified as Top Secret; neither do they contain entries if the individual entry itself contains classified information. Appellant's Br. at 9.

On the basis of his review of four TABs which have been released, AFSC's expert opines that "very few reports are summarized." Johnson Aff. at 2 par. 5, appendix at A-152. But one of DoD's experts, who reviewed all of the TABS generated in that year, appears to say that 74% of all entries in 1983 TABs contain abstracts. Robey Aff. at par. 4, appendix at A-148. We return below to this difference of opinion.

II. WITHHOLDING TO PROTECT NATIONAL SECURITY:

The national security exemption is invoked here on the theory that, while the individual entries are not classified or (therefore) withholdable, the aggregation is withholdable on the "compilation" theory.

The national security exemption concerns matters that are

A) specifically authorized under criteria established by an Executive order to be kept secret in the interest of national defense or foreign policy and (B) are in fact properly classified pursuant to such Executive order.

5 U.S.C. Sec. 552(b)(1) (1982). In conformity with this requirement, Executive Order 12,356 provides for the classification of information which "either by itself, or in the context of other information reasonably could be expected to cause damage to national security." While AFSC does not contest the validity of the compilation theory,1 it does argue that DoD is obliged to produce those portions of the requested material which can be segregated from the information whose disclosure would damage the national security. Apart from its segregability contention, AFSC also contends that in light of the factual record, the compilation theory does not apply here because DoD's judgment that national security "reasonably could be expected to" be damaged cannot be respected.

B. Deference Owed DoD's Judgment On Damage to National Security

The standard to be applied in assessing DoD's decision to withhold has been described as follows by the District of Columbia Circuit:

Because " '[e]xecutive departments responsible for national defense and foreign policy matters have unique insights into what adverse affects [sic] might occur as a result of public disclosure,' " however, courts are required to " 'accord substantial weight to an agency's affidavit concerning the details of the classified status of a disputed record.' " Salisbury v. United States, 690 F.2d 966, 970 (D.C.Cir.1982) (quoting S.Rep. No. 1200, 93d Cong. 2d Sess. 12 (1974), U.S.Code Cong. & Admin.News, 1974 pp. 6267, 6290). Accordingly, an agency is entitled to summary judgment if its affidavits "describe the withheld information and the justification for withholding with reasonable specificity, demonstrating a logical connection between the information and the claimed exemption," id. at 970, and "are not controverted by either contrary evidence in the record nor by evidence of agency bad faith." Military Audit Project v. Casey, 656 F.2d 724, 738 (D.C.Cir.1981).

Abbotts v. Nuclear Regulatory Commission, 766 F.2d 604, 606 (D.C.Cir.1985). We adopt that standard here.

DoD's concern here is that, by looking at all of the TAB entries, a person could draw useful inferences about the direction of U.S. research into weapons or defensive technologies. AFSC responds that, when a project begins to approach fruition it will stop appearing in the TABs, as it will if DoD concludes that the project is not worthy of further exploration. DoD acknowledges this, but contends that the disappearance of a given project from the TABs will still give those seeking such information help in ascertaining what projects should be the target for their own intelligence work.

one would have no way of determining whether [the subject's disappearance from the TABs] was because research had been terminated (or for what reasons), or because it had reached fruition, so that further reports were classified "Top Secret" and/or their titles were themselves classified.

Appellant's Br. at 11.

We do not find this a satisfactory rejoinder. Even if a hostile nation could not tell which of the two alternatives was correct, it would be able to infer that one of them was. This would enable it to concentrate its intelligence resources on those areas which were likely to be most important.

AFSC correctly observes that, in response to an FOIA request, an agency must disclose segregable portions of otherwise nondisclosable material. See Founding Church of Scientology v. Bell, 603 F.2d 945 (D.C.Cir.1979). AFSC goes on to argue, however, that the segregability doctrine requires DoD to disclose all material that is not itself classified. Many of the titles contained in the TABS are not classified, and AFSC concludes that DoD must disclose those titles. We believe that acceptance of the compilation theory requires that this argument be rejected.

By invoking the compilation theory, DoD argues that material which could not harm the national security when disclosed by itself can be withheld if its disclosure--along with other material whose disclosure alone would also not harm the national security --might endanger the national security. See Halperin, 452 F. Supp. at 50. AFSC's insistence that everything which is not classified must be disclosed essentially ignores the compilation theory because it ignores the danger to which the compilation theory points--that information harmless in itself might be harmful when disclosed in context. The fact that the material withheld is not itself classified does not in turn require disclosure if the compilation theory is invoked, because the theory itself is only relevant when the material withheld is not classified. If the material were classified it could be withheld without reference to the compilation theory.

The compilation theory can be based on a number of different problems which might be caused by the publication of unclassified data. Neither plaintiff nor defendant has explored this issue, but we find it useful to separate out the problems we can think of in order to explain our holding and to give direction to the district court on remand. While we attempt here to define the different problems to which the compilation theory is a response, we are far from certain that our list is exhaustive.

First, as Halperin suggests, the sequence in which documents were generated might reveal damaging information which the documents themselves do not disclose. Revealing the whole series, therefore, complete with the dates on which the individual documents were written, might be damaging even though disclosure of any one (or perhaps two) documents would not be. The doctrine of segregability suggests, however, that if this is the problem DoD is worried about we should respond to it by deleting the dates of the reports, or by otherwise making it impossible for a reader of the information disclosed to determine the order in which individual reports were written.

Second, as the parenthetical in the last paragraph suggests, there is a "slippery slope" problem lurking in the background of the compilation theory. Disclosure of one entry in a TAB would likely not do significant damage to the national security, and the same would seem to be true for two entries, and for three. At some point, however, the compilation theory suggests, the number of entries becomes so large that a viewer could arrange the entries to get a picture of the nation's national defense research program, much as a jigsaw puzzle's picture can be guessed although not all the pieces are in place. The doctrine of segregability suggests that we should order the release of that number of reports which can be disclosed without the whole picture becoming "guessable." We are most reluctant to determine, however--as a matter of law, which is how the issue must eventually be resolved--what the number of entries is that at which the picture becomes "guessable." That judgment, it seems to us, is within DoD's expertise. DoD has exercised that expertise, via its proffered affidavit, in a way we do not find implausible or unreasonable, and we will not disturb its conclusion.

We note that the affidavit of AFSC's expert, Johnson, states that

"a significant number of the entries in the TAB also appears in substantially the same form in the National Technical Information Service catalog, Government Reports Announcements/Entries, which is available at public libraries across the United States."

App. at 153. It is not clear what is meant by a "significant" number of entries. Nor is it clear from the affidavit in what form the entries are set forth in the public catalog.2 At all events this conundrum points to the existence of a genuine issue of material fact and hence raises a question as to the propriety of summary judgment.3 We identify an even more serious problem in the next section of this opinion.

Finally, as the last possible problem to which the compilation theory points, DoD argues that a knowledgeable reviewer of the TABs could make valuable and damaging inferences from the presence in the TABs of reports on related topics. This depends, of course, on how much information each of the TAB entries contains. That depends in turn on whether the TAB entries contain summaries of the reports--a question which we are unable to answer on the record as it now stands. If the TABs do contain abstracts, and if they are the parts of the TAB entries which convey damaging information, then perhaps the TABs could be disclosed after the abstracts had been deleted. On the other hand, it is possible that even the titles of the reports contain enough damaging information that their disclosure--without a summary of the report--would harm the national security. We remand so that the district court can determine which of these statements is correct.

AFSC's expert stated in affidavit that "very few reports [whose titles are contained in TABs] are summarized," and that "most TAB entries don't even contain a full abstract. Most include just a title, dates, report numbers, descriptors and other details that are not likely to shed any light on the details of the report itself." Affidavit of David T. Johnson, Director of Research, Center for Defense Information, at 2, 4-5, Appendix at A-152, A-155-56 (emphasis added). This testimony was based on Johnson's review of four of the 26 TABs released during 1983 (the first year for which AFSC has requested all TABs).4

DoD's expert Paul Robey, in contrast, gave significantly different information on the number of TABs containing abstracts. Robey, who is Deputy Administrator of the Defense Technical Information Center (DTIC), provided the following information in his affidavit:

DTIC does not keep archival copies of individual TAB issues, therefore, an exact or average number of abstracts is not available. In 1983, DTIC ran a computer review of the beginning and ending sequence identifying numbers assigned to the technical reports announced in the 1983 TABs. The computer identified 12,334 numbers within the range representing 12,334 technical reports or entries. Of the 12,334 entries, 9,083 or approximately 74% included an abstract. The file from which these numbers were derived changes periodically and the numbers given here may not reflect what was originally published in the 1983 TABs. Some of the identifying numbers, known as AD numbers, may have been cancelled or replaced.

App. at 148-149.

We are unable to see how the statements by AFSC's and DoD's expert witnesses can be deemed consistent with one another. AFSC's witness says that only "very few" TAB entries contain abstracts. DoD's witness implies that 74% of them do so, but he actually says only that 74% of the entries in the file contain abstracts. He does not say that the entries appear in the TABs in the same form as they appear in the file upon which he bases his testimony; as noted above, the affidavit is based upon the file, not the TABs of which DTIC, Robey's agency, did not have archival copies. We are therefore not certain of the number of TABs Robey is claiming contain abstracts. But the clear implication of his affidavit is that 74% of them do so. Summary judgment on defendant's behalf can be sustained, therefore, only if this difference between the two experts is immaterial.

As the quotation from Abbotts makes clear, see supra p. 434, FOIA's legislative history instructs the courts to defer to DoD's judgment that the national security would be endangered by the disclosure sought here. That judgment is based on information, experience and expertise to which we are not privy, and we will not lightly override it on the basis of our own understanding of what does or does not constitute a danger to the country's security. In light of the deference to which DoD is entitled, we may be obliged to respect its judgment about the danger of disclosure even if most of the TABs do not contain abstracts. We believe, however, that it is unnecessary for us to decide that issue now.

Because the record is unclear at this stage as to the fraction of TABs containing abstracts, we have some difficulty concluding that there is a significant possibility that the TABs' disclosure will indeed harm the national security. We would be more comfortable deferring to DoD's judgment on this point if the record showed clearly--or if there were a factfinding--that a significant percentage of the TABs contain abstracts. If that were the case it would be much easier to understand why DoD has exercised its judgment as it has, and we would doubtless be obliged to affirm the judgment on DoD's behalf. We will therefore vacate the grant of summary judgment and remand to the district court to permit resolution of the disagreement between plaintiff's and defendant's expert on the issue of the percentage of TABs containing abstracts, and if it appears that it is small, what danger there may be of disclosure in that event.5 See also discussion supra at p. 446 & nn. 2, 3.

Remand would be unnecessary, however, if the other FOIA exemption invoked by DoD will justify withholding. The alternative ground invoked by DoD is 5 U.S.C. Sec. 552(b)(3) (1982). That provision exempts (from FOIA) information that is

specifically exempted from disclosure by statute (other than section 552b of this title), provided that such statute ... (B) establishes particular criteria for withholding or refers to particular types of matters to be withheld.

DoD argues that 10 U.S.C.A. Sec. 130(a) (West Supp.1987) is such a statute in that it confers on the Secretary of Defense the express authority towithhold from public disclosure any technical data with military or space application in the possession of, or under the control of, the Department of Defense, if such data may not be exported without an approval, authorization or license under the ... Arms Export Control Act."

"Technical data" are defined in section 130(b)(2) as

any blueprints, drawings, plans, instructions, computer software and documentation, or other technical information, that can be used, or be adapted for use, to design engineer, produce, manufacture, operate, repair, overhaul, or reproduce any military or space equipment or technology concerning such equipment.

DoD argues that the TABs contain "technical information[ ] that can be used" in the above-described way. Whether that is true, however, turns on whether or not the TABs actually convey any of the contents of the reports whose titles they list. That is the same question as the one upon which the applicability of the national security exemption depends. We decline on this record to assess the applicability of the "authorized withholding" exemption for the same reason that we have declined at this time to decide the applicability of the national security exemption.

The judgment of the district court will therefore be vacated and the case remanded for further proceedings consistent with this opinion.

I disagree with the disposition ordered by the majority which remands this case to the district court for factfinding concerning the number of TABs which contain abstracts.

The majority opinion clearly holds that the mere compilation of titles1 provides an opportunity for the disclosure of classified information and that the Department of Defense's decision to refuse disclosure was reasonable. Maj. op. pp. 445-446. If the opinion had stopped at that point, I would have no difficulty in joining the opinion and its holding.

However, despite reaching this conclusion, the majority then goes on to require a remand so that the district court may reconcile or determine the percentage of TABs containing abstracts because the parties' expert witnesses differed with respect to this fact. I cannot fathom why or how this information can bear on the decision in this case which rests on a compilation theory.

Once we have decided that the mere compilation of titles without more may reveal sensitive information, the issue has been concluded. Indeed, the majority opinion states as much: "In light of the deference to which DoD is entitled, we may be obliged to respect its judgment about the danger of disclosure even if most of the TABs do not contain abstracts." Maj. op. at 447. Thus, whether or not the titles are accompanied by abstracts, is not only irrelevant, but in light of the majority's remand, it calls for factfinding in a summary judgment context that is clearly inappropriate.

Because I am satisfied with the majority's compilation theory analysis and its reasons for deferring to the Department of Defense's judgment, I would affirm the order of the district court which granted summary judgment in favor of the Department of Defense.