Rabbit's Warren

This is Part 2 of a series of posts I am doing on CIA records in the ARC which were designated “not believed relevant” (NBR). This post corrects the NBR count for CIA documents given in Part I, examines the release history of these NBR documents, and begins a discussion of ARRB memos on groups of NBR records.

I should note here that records from agencies other than the CIA were also either designated NBR or treated in an analogous way. I will discuss these records in a future series of posts. This series will discuss only CIA records.

Introduction: ARRB and NBR

The Assassination Records Review Board (ARRB) was an independent federal agency established by Congress to oversee the creation of the JFK Assassination Records Collection (ARC). Many of the documents placed in the Collection were redacted by the originating agencies on national security, law enforcement, and privacy grounds.

It was the ARRB’s responsibility to review these redactions, opening as much of the documents as possible to the public. This was a lengthy process that often involved extensive consultations with the orginating agencies.

As the ARRB began work on the CIA “sequestered collection” (SC), a massive set of documents numbering close to 300,000 pages, it became concerned that some of the material in the SC was of “marginal relevance” to the JFK assassination, and that a “word by word” review of redactions to such materials would curtail its work in other important areas.

To reduce the time spent on marginal material, the ARRB developed a set of guidelines for processing material “not believed relevant” (NBR) to the assassination. If a page by page review of a document found it of no relevance or marginal relevance, the Board designated it NBR and deferred release of the entire document until the October 2017 deadline for release of all materials in the ARC.

Revised count of CIA documents declared NBR

As my previous post on the NBR material noted, it is sometimes difficult to tell which records were in fact designated NBR. My primary basis for distinguishing NBR records is the Assassination Collection Reference System (ACRS), a database of metadata for ARC records provided online by the National Archives and Record Administration (NARA), which holds all the ARC records.

However, as the previous post also noted, there are some cases where the ACRS metadata does not indicate a record was designated NBR, but other ARRB material indicates that it was. For the CIA documents, these “other materials” were mostly notices of record determinations which the ARRB published in the Federal Register.¹ In my previous post on NBR records, I found 762 NBR records for the CIA in the ACRS database, and 29 notices of NBR records in the Federal Register which were not included in the ACRS.

After review, however, I have revised these figure: I now count 42 records as NBR using data from the ARRB’s notices in the Federal Register, and 758 records as NBR using the data in the ACRS. I missed the ACRS records before because they use slightly different language in describing the NBR designation. I identify the additional records in the Federal Register because they were noticed as postponed in full, which the ARRB only did for CIA records designated NBR. In addition, the ACRS comment fields for records gives the same description used for the other Federal Register NBR records.

If all of these revisions are correct, that means the total number of CIA NBR records is 800. I have posted a new excel file that lists these records here.²

Release of the NBR documents

As far as I can tell, the CIA documents designated NBR were withheld in full until their release in 2017-2018. There are several reasons for thinking this is so.

First, I have found no trace of these documents in the Mary Ferrell Foundation collection of ARC documents. The MFF collection is of course not exhaustive, but the greatest strength of the MFF collection of ARC documents is its set of CIA records. The fact no NBR CIA records were in the MFF collection (until they were released in 2017-2018) certainly suggests that these documents were not released prior to 2017-2018.

Second, virtually all of the NBR CIA documents were on the NF16 list. This was a list of all ARC records supposedly withheld in full, released by NARA in response to an FOIA request. As I have noted elsewhere, not all of the records on NF16 were actually withheld in full; some of these documents are present in other places in the ARC collection. I have found no such traces,however, of any NBR records³.

Third, the finding aid attached to each ARC document, known as a RIF sheet, lists the status of the record: open in full (OIF), withheld in part (WIP), or withheld in full (WIF). I have found no RIF sheets for the CIA NBR documents that indicate anything other than WIF (or sometimes DIF “Denied in Full”).⁴

Finally, ALL of the NBR records were “released” in 2017-2018. This is not an absolute guarantee that all of these documents were previously withheld in full. As I have also noted elsewhere, “released” in the context of the ARC is a technical term, and simply means that previously redacted text in a document has been opened to the public. Thus a 500 page document that was withheld in full by CIA in 1993, then released by the ARRB in 1996 with only two letters redacted, would still count as a “release” if the final two redacted letters were made public in 2017. Similar instances happened with a number of records in the 2017-2018 releases. It is therefore possible that some of the NBR records “released” in 2017-2018 were not released for the first time, but had also had text released earlier. I have found no evidence of this, however.

Summing up, I think it is very likely that ALL of the CIA NBR records listed on my excel sheet were in fact withheld in full until released in 2017-2018.

NBR document groups

Although determining whether specific records were designated NBR is sometimes a problem, ARRB files include a number of memos on groups of records (also called record blocks or sets) that were designated NBR. Many of these memos were written by Michelle Combs, who was ARRB’s Associate Director of Research and Analysis when the Board ended in September 1998. These memos are available online at Mary Ferrell (here). A good overview of the main sets of NBR documents can be found in this memo by Combs, which lists the following groups of NBR documents:

• Nosenko records: Yuri Nosenko was a KGB defector who claimed to have knowledge of the KGB’s records on Lee Harvey Oswald. His bona fides were the subject of great controversy inside the CIA. The ARRB found a number of Nosenko documents relevant to the assassination and released these in 1995-1998. The NBR Nosenko documents were not released until 2017-2018. Combs estimates these totaled 2,400 pages.
• CRC financial records: ARRB staffer Manuel Legaspi describes the Cuban Revolutionary Council (CRC) as “an umbrella group of anti-Castro groups formed with the support of the U.S. Government, [that] was to be the basis for an official government of Cuba had the Bay of Pigs invasion been successful in ousting Castro from power.”⁵ This group of documents contains “detailed financial and monthly accounting records” of the CRC. Combs estimates 5,400 pages of documents in this group.
• Office of Personnel Files: According to Combs, ‘The HSCA was provided with the official Office of Personnel (OP) files for every agency employee connected in any possible way with the assassination or any investigation of the assassination.’ Combs does not give a page estimate for this group.
• The Monster NBR: This was a list compiled by the ARRB’s CIA team. After a review of all the microfilm material in the CIA’s “Sequestered Collection” which the CIA designated non-relevant, the ARRB reviewers came up with their own list of materials which they agreed were NBR. This list was apparently classified at the time of its compilation, and I have not seen a copy of it. Combs estimates 25-30,000 pages of documents on the Monster list were marked NBR.

Coming next

The next several posts in this series will cover these four groups of NBR documents. This is worth doing for at least two reasons. First, many of these records have their own historical value. The debate over Nosenko’s bona fides, for example, has been the subject of several books, and I believe the newly released NBR documents shed significant light on the subject. Second, we can now use the released records to independently evaluate the ARRB’s decision to designate these documents as not relevant to the JFK assassination.

[This post was revised on 5/7/2019 to add the endnotes that I accidentally omitted when posting, and to fix a few typos. Unfortunately, the count of NBR documents is also slightly off, and will be revised in the next post.]

From 2017 to 2018, the National Archives and Records Administration (NARA) released seven sets of documents from the John F. Kennedy Assassination Records Collection (ARC). News media and assassination researchers have claimed these releases include significant new information about the assassination of JFK, previously withheld from the public.

In a recent post I looked at responses to such claims by Max Holland and Dale Myers, two writers who have done important work on the JFK assassination. As both made clear, the significance of the releases has often been overstated.

This post discusses a set of documents from the 2017-2018 releases which received much attention, but whose signficance is particularly questionable. These are the CIA files designated “Not believed relevant” to the assassination (NBR) by the Assassination Records Review Board (ARRB). The ARRB was the independent federal agency charged with overseeing the collection and release of government documents on the assassination of JFK.

In this post I look at why the ARRB designated some records as NBR, the criteria it followed when doing this, and the problems I had in identifying these records. For those who want to follow the discussion in detail, all of the figures and records cited below are listed in an excel sheet posted here. In future posts I will look at specific topics in CIA records with “no believed relevance” to the JFK assassination. I will also take a look at a similar, even larger set of FBI records.

The creation of the NBR designation

The genesis of the ARRB’s NBR designation was at an ARRB meeting on August 6, 1996.¹ The ARRB’s first year and a half of work on CIA records was devoted to a word by word review of the CIA’s main file on JFK assassin Lee Oswald, usually referred to as the Oswald 201 file. By August 1996, the Board’s work on the Oswald 201 file was done, but the CIA still had a huge number of records that the ARRB needed to review for relevance to the JFK assassination.

The bulk of these records was in the CIA’s “sequestered collection” (SC), sometimes called the segregated collection. This collection consisted of all the records the CIA had gathered for the investigation conducted by the House Special Committee on Assassinations (HSCA).

The transcript of the August 6 ARRB meeting includes an introduction to the sequestered collection, given under oath by two members of the CIA’s Historical Research Group (HRG): John Pereira and Barry Harrelson. Pereira’s presentation divided the SC into two parts: hard copy files with about 129,000 pages, and 72 reels of microfilm, equivalent to about 163,000 pages when printed out, so a total of about 300,000 pages.²

The Board members believed that processing such a large number of records using the exacting word by word review standards they employed in the Oswald 201 file would consume most of the ARRB’s remaining time.³ As Board member Anna Nelson observed during the August 6 meeting:

One of the reasons we feel — and we’ve talked about that we feel this is such a big problem — is that if we take the time to word-for-word look at the sequestered collection, then we won’t have time for anything else.⁴

The Board was also concerned that much of the material in the SC was of secondary value. Many documents were duplicated, sometimes massively. In the CIA presentation, Pereira cited one document which the CIA reviewers had found 43 copies of.⁵

In another presentation at the same meeting, T. Jeremy Gunn, ARRB counsel and associate director for research and analysis, also notes that in the SC “there are some records where it is very difficult to determine the relevancy [to the assassination].” In fact, as Gunn’s presentation makes clear, there are large chunks of the SC that are of no discernible relevance at all.⁶ I will have a detailed discussion of this problem in my next post on the NBR records.

Following Gunn’s presentation, the Board also provided time for comments from JFK researchers, including authors John Newman and Harrison Livingstone, AARC president Jim Lesar, and COPA president John Judge. Newman, Lesar, and Judge all argued that the Board should aim to fulfill both goals: complete processing of the SC, and continuing to search for and process new documents. If necessary, they suggested, the board could ask Congress for a further extension of its term.⁷

Board member Anna Nelson expressed doubt that further extension of the Board’s term would pass Congress and pressed Lesar on his priorities:

MS. NELSON: For a moment, let’s forget about an extension, which would be difficult in this Congress. Would you, Mr. Lesar, prefer that we examine the new file materials and not seek any other CIA or other documents?
MR. LESAR: No. Given that choice, I would prefer that the board give primary attention to drawing unidentified assassination records into the collection.

As the transcript of the August 6 meeting shows, this sentiment was shared by the Board. On the other hand, the Board’s reading of the ARCA’s requirements made them reluctant to consider non-release of the irrelevant material in the SC (this is discussed below).

The NBR guidelines

Caught between the two imperatives of continuing the search for more unidentified assassination records and full processing of all material in the SC, the ARRB came up with a compromise solution for handling documents of marginal or undetermined relevance in the SC. This solution was implemented in a set of guidelines which ARRB staff began drafting soon after the August 6 meeting.⁸ The final version of the guidelines was adopted at the Board meeting on 13 November 1996.⁹

These guidelines had not been drafted when the ARRB’s 1996 report was written, and they are only briefly mentioned in the ARRB final report,¹⁰, so a more detailed look at them is appropriate here.

The guidelines first defined a set of “segregated collections” consisting of FBI and CIA records provided to all government investigations of the JFK assassination. The guidelines require a complete review of all records in these collections. Records that are relevant to the assassination, or that “enhance historical understanding” of the assassination, were to have all requests for “postponements” (redactions) reviewed on a word by word basis.

Records that the ARRB staff decided had information “not believed relevant” (NBR) to the assassination were to be documented with a brief description of the basis for the determination. The Board would then stopped processing the document or the section of the document determined to be NBR. The effect of this was that these documents, for the most part, were withheld in full through the ARCA’s 25 year deadline, though the ARRB encouraged the agencies to release the information.

The guidelines provided that the record’s final determination form “shall reflect that such postponements have been sustained on both the specific grounds enumerated in Section 6 and the material’s NBR status.” and emphasized that “Under no circumstance shall information that is relevant to the assassination be postponed on joint NBR-Section 6 grounds.” Put more plainly, these rules did not allow withholding any information in the NBR documents that was relevant to the JFK assassination.

Problems in NBR record accounting

Accounting for which records were NBR proved to be a problem for the ARRB. The most complete list of NBR records that I have found is in the Assassination Collection Reference System (ACRS), NARA’s online database of ARC metadata. The ACRS lists 762 CIA records as NBR. These are records where the phrase “NBR” appears in either the subject field or the comment field of the record. I have listed the titles, RIF #s, subjects, and comments for the records I found in the excel sheet for this post (here).

To ensure that the ARRB’s review of assassination records was performed with a high level of public scrutiny, the ARCA also required the ARRB to publish notices in the Federal Register for all record decisions.¹¹ The NBR guidelines, however, did not specifically deal with publication of NBR determinations in the Register. In fact, only 55 NBR record notices were published in the Register. These are also listed in the excel file for this post (here).

Of these 55 NBR records, only 33 were from the CIA. Of the 33 CIA records listed as NBR in the Federal Register, only four are listed as NBR in the ACRS. What happened to the other 29? These records are all from CIA disk number 104-10063, with records numbered 104-10063-100XX. Checking the full metadata for these records, they all have a note in the comment field that reads “FBI DENIED IN FULL 2/16/94; ARRB DENIED IN FULL 3/14/97.” This is instead of simply writing “NBR” as all the other NBR records do.

The metadata for all CIA records in the ARC was keyed in by the CIA reviewing team, so this confused, inaccurate description is on them. In defense of the review team, we should also note that these were the first records that the ARRB declared NBR, so the review team perhaps did not immediately understand on what basis the documents were withheld.

Adding the 762 NBR records from the CIA in the ACRS, and the 29 records listed as NBR in the Federal register gives a total of 791.

Why has there been confusion about these records? First, with the exceptions noted above, they were not noticed in the Federal Register, in contrast to all the other records processed by the ARRB. Second, with one or two exceptions, the “final determination forms” which the NBR guidelines envisioned would be attached to all of these documents were apparently never prepared. Third, the staff documentation for the NBR records (“a brief description of the basis for the [NBR] determination”) was not published, but remained in the ARRB files. I am sure that it exists there, but no researcher has ever bothered to dig it out. Documentation for some of the classified files is also probably still classified itself.

Relevance in the Assassination Records Collection

Ultimately, as far as I have been able to trace them, the NBR records from the CIA were all included in the ARC, and have been released, either in full or in part. To me, this poses something of a puzzle. If these records are truly unrelated to the JFK assassination, why did the ARRB insist on including them in the collection?

The Board’s primary rationale for including such irrelevant material in the ARC is that it was obtained by the HSCA as part of its investigation, and Section 3(2) of the Assassination Records Collection Act (ARCA), the legislation establishing the ARC, defines an assassination record as “a record that is related to the assassination of President John F. Kennedy that was created or made available for use by, obtained by, or otherwise came into the possession of [the government].”¹²

NF16 is a list of 3598 records in the JFK Assassination Records Collection (ARC) at the National Archives and Records Administration (NARA).¹

NARA released the list in January 2016 in response to an FOIA request by Michael Raznitsky, who asked for a list of all records in the ARC “withheld in full” (WIF) as of November 2015.²

As discussed in earlier posts, WIF is one of three possible states for ARC records. The other two are “withheld in part” (WIP), and “open in full” (OIF). OIF records have no text removed; they are “unredacted”, and thus completely open to the public. WIP records still have some text removed for reasons of national security, law enforcement, or privacy. They are thus “redacted.” WIF records are records that are not available to the public at all, except for certain items of metadata relating to the document, i.e. number of pages, basis for withholding, agencies which provided the document, and so on.

WIF records have been a focus of interest for many people interested in the JFK assassination, but for reasons that I will discuss in another post, the actual number of WIF records in the ARC was left in confusion when the main work on declassifying and opening the records ended in 1998. NF16 was part of NARA’s effort to clarify this confusion.

My post today takes a second look at the NF16 list, in an attempt to see how many documents on the list were ultimately released, problems it revealed with NARA’s ability to identify WIF records in the ARC, and researchers’ sometimes flawed understanding of what is available and what is not.

NF18 and NARA 18

Before I look back at NF16, however, a review of two other lists relevant to the 2017-2018 ARC releases is necessary. These lists include NF18³ and NARA18.⁴

NF18 is a list of 22,933 records in the ARC. The NF18 list was released in January 2018 in response to another FOIA request, this one from John Greenewald, who runs The Black Vault website.⁵ Unlike NF16, which lists only WIF records, NF18 was supposed to list all records in the ARC which still had redactions as of the date of its release. The NF18 list has many duplicate records, so for purposes of counting or comparison total redactions, one must use the number of unique records in the list, which is 21,890.

NARA18 is the cumulative list of all seven ARC record releases from July 2017 to April 2018. It gives the final tally of ARC records released in this period. Note that “record releases” here is a technical term, it does not mean a complete restoration of all redacted text at one fell swoop, but instead refers to a document that has had at least one redacted passage restored. In fact, some documents were “released” multiple times, with different text redactions restored at different times. Tracking what text was restored in which release is a permanent source of confusion in any accounting of ARC documents.

Of the 21,890 records on NF18, 798 are identified as “withheld,” meaning in this case “withheld in full.” The remaining 21,092 records are identified as “redacted,” meaning in this case “withheld in part.” In the most recent release of ARC records on April 26 2018, NARA clarified the status of the “withheld” records, most of which are not eligible for release under the 1992 Assassination Records Collection Act (ARCA) the law governing the ARC. Only 9 of the 798 files were released in April. In addition to the ineligible files, another 200 or so were record errors, damaged recording tapes, and a large set of microfilmed duplicate files on Lee Harvey Oswald, the originals of which had already been released earlier.⁶

Adding all these together, NARA has now accounted for all the WIF files in ARC. The files in the ARC which remain WIF will remain closed to the public, barring a change in the ARCA, the law which authorized releases from the Collection. How many of the WIP files still in the Collection will remain redacted is unclear (I will do a post on this in the near future). There is also some uncertainty as to which records in the ARC are now OIF and which remain WIP. Eventually one would hope NARA will clarify these issues, but this will probably not happen in the near future.

A comparison of NF16, NF18 and NARA18

Going back now to NF16, one must remember that this list was compiled over 3 years ago. Although it was originally supposed to be a list of only the WIF files in the ARC, it is now clear that a number of the files in the list were in fact NOT withheld in full. Nonetheless, 797 of the 798 withheld in full documents in NF18 were already listed in NF16⁷

Comparing the remaining files, 2447 records from NF16 appear in NARA18. These are all releases of previously redacted material. I should note that it was probably not the case that all of these documents were actually WIF, but even if they were WIP at the time of the 2017 releases, we at least have the documents and can now examine its contents.

In any case, 2447, plus the 798 WIF unreleased records listed in NF18, still leaves something over three hundred records in NF16 unaccounted for. These unaccounted records were noted by several websites which have been tracking the releases. Jimmy Falls noted this gap at the WhoWhatWhy.com website soon after the release of NF18,⁸, and Rex Bradford, president of the Mary Ferrell Foundation made the same point in a June 2018 overview of the 2017-2018 releases.⁹

Fortunately, Falls provided a copy of the NF16 pdf with the “missing” records highlighted in green, so it is possible to immediately identify 374 of the 375 records he is interested in.¹⁰ Bradford agrees with Falls that there are 375 files in NF16 “which had not been released in 2017” and are not present in NF18. The Mary Ferrell Foundation corresponded with NARA on this question, and according to Bradford “the Archives in a reply claimed that 336 of the 375 had been erroneously included on the 2016 list in the first place. The other 39 were said to be ‘pending April release’ despite being missing from the 2018 listing.”

I believe I have now tracked down these discrepancies, but the one by one itemizing was a truly dull task. I will put up an excel sheet with the items when I have recovered from my Stakhovite labors. In the meantime, I will give a short version here.

First, there are actually more than 375 documents “missing” from the NARA18 spreadsheet. NF16 as originally released was missing the final page. The Mary Ferrell Foundation discovered this omission and got the missing data from NARA.¹¹ Falls’ copy of NF16 is missing this final page, which lists an additional 27 documents. 18 of these also missing from NF18. Like Falls, Bradford counts 375 “missing documents” in both the “Open Letter” and his report on the 2017-2018 ARC releases, so he has also omitted the 27 documents on the final page of NF16.

Neither Falls nor Bradford mentions actually trying to look up the 375 records they question at NARA. I have not myself checked all 336 records which NARA told Bradford were included in NF16 by error, but just glancing at other documents online at NARA or Mary Ferrell show several cases where these records were incorporated in other materials, and are present in full, with no redactions. As examples, record 179-40003-10035 appears in Commission Document 442, starting on page 74. This document is available on Mary Ferrell (see here). Record 179-40005-10139 appears in the same document, starting on page 78. (see here)

Comparing the RIF metadata sheets for 179-40003-10035 in CD442 with the data in the Assassination Collection Reference System (ACRS), NARA’s online database of ARC finding-aids, shows that 179-40003-10035 was reviewed on 9/14/94, and that when its RIF sheet was printed, it was OIF. Yet its entry in the ACRS was never updated to reflect this change. This seems to have occurred frequently in documents with the prefix 179; over half of the 336 documents erroneously listed in NF16 are in fact documents with the 179 prefix. This failure to update the ACRS is perhaps the main reason why NARA has struggled to identify which records are WIF, WIP, and OIF.

While this is a problem which needs correction, I find the critiques of both Falls and Bradford on NF16 off the mark. Falls did not even bother to correspond with NARA on the question of the “missing” documents. Bradford did, but still apparently questions NARA’s response that 336 records released in full were incorrectly included on NF16, without checking the actual state of the documents at NARA, and without bothering to check NARA18 to see whether it included 39 other “missing” documents. (Or if he did, he did not bother to inform us that they were indeed there.)

To question NARA’s ARC figures without taking the most basic step of actually looking up the records in question shows a lack of concern for accuracy. At this late point in the development of the ARC, when we are not talking about vast numbers of documents but a few thousand, a few hundred, or even a dozen or so, there is no excuse for not doing this.

NARA is not obligated to be 100 percent correct in every metadata item for every document in the ARC, just as the Library of Congress is not obligated to be 100 percent correct. It should show due diligence in revising errors, and it should give due notice of where errors may lie. In the case of NF16, prior to its release, NARA more than once reminded researchers that its database field for record status (OIF, WIP, WIF) was NOT accurate.¹² It revised lists of documents and responded to researchers’ questions. As far as I can see, NARA has made due efforts to open all records that the ARCA allows to be open.

NARA has been less prompt in updating and correcting the ACRS. According to NARA staff member Gene Morris (in a recent email to me), this is a problem of funding. Given the vast amount of time and resources poured into the ARC, however, it makes no sense to leave the ACRS in its current condition. It should be updated and corrected to provide complete and accurate metadata on all records in the ARC, and I have written to my representatives to ask that they consider fully funding this final step in fulfilling the goals of the ARCA.