Rabbit's Warren

For those not familiar with the subject of this post, I have a new page introducing the JFK Assassination Records Collection (a link is at the top of the blog page). This should help make the discussion a little more comprehensible.

I’ve now had a look at the two reports I mentioned in my last post. These reports cover the final releases of material from the JFK Assassination Records Collection (ARC), which is held at the National Archives and Record Administration (NARA). Neither report is as useful as I had hoped.

The first of the two reports is from the office of NARA Inspector-General James Springs (here). It was issued 29 March 2018, a little less than a month before NARA released what is supposed to be the final set of ARC documents. The IG report therefore only covers the 2017 ARC releases.

I should note that I am not certain whether the document I cite here is the report itself, or simply a summary of the report. ¹

Much of the report is simply a recitation of the basic history of the JFK assassination; no need to go over that here.

The remaining part of the report recounts various aspects of the 1992 ARCA, the legislation which created the ARC. It notes that, in the IG’s view, the ARCA is still in effect and imposes a “continuing obligation” on federal agencies to release assassination records.

It also spells out the records which the ARCA exempted from public release, including sealed court documents, federal grand jury information, deeded gifts to the federal government, and “records held under section 6103 of the Internal Revenue Code.”

There is then a brief description of NARA work on the ARC releases, beginning in 2014 and ending with work current as of the report date (March 2018).

For those who have not followed this long story, in 2014 the JFK Assassination Record Collection still had numerous records that were not available to the public, or available only in part.

The ARCA set a 2017 deadline to release most of this information, so after lengthy preparation and consultation, described in the IG report, NARA released six sets of records from the ARC during July to December 2017. According to the IG report, during this period 34,873 documents were ‘released.’

A caveat is now in order here for readers who may be confused by the word released.

Many readers have taken the word “released” in this context to mean that a document was previously “withheld,” i.e. unavailable to the public. This is not how these terms are used in describing ARC releases.

There are three possible states in which an ARC record may appear: withheld in full, withheld in part, and released in full. Documents withheld in full are not available to the public at all. Documents released in full are publicly available, with no text removed at all. Documents withheld in part are publicly available, but have part of their text removed. These removals (usually called ‘redactions’) range from a single excision of as few as two letters (a digraph), to several pages of text.²

When a redaction is “removed” from a document, this means the blanked out or blacked out text is restored. For ARC documents, however, the restored text is said to be ‘released.’

Why is this confusing? Consider a hypothetical ten page document, published ten years ago with only one redaction, two letters from a CIA cryptonym. When the two letters are restored, NARA counts the entire document as a “release.” Moreover, if redaction x in a document is ‘released’, but redactions y and z are not, the same document may later be referred to as ‘released’ again and again, until it finally reaches the stage of ‘released in full.’

This counter-intuitive terminology led some writers last year to believe that ALL of the documents ‘released’ during 2017 were previously withheld in full! Since a total of 34,873 documents were ‘released’ in 2017, it is easy to get the mistaken impression that a vast amount of material was previously unavailable to the public, based on the false assumption that all of these 34,873 documents were withheld in full.

But this was far from the case. Pursuant to an FOIA request in 2016, NARA attempted to compile a list of all the withheld in full documents. This list, which I refer to as NF16, counted 3598 documents withheld in full.³

The vast majority of the 34,873 documents released in 2017 were thus NOT withheld in full. Many of them had only one or two redactions, and after the 2017 ‘release’ were actually ‘released in full.’

Ah, but which ones? Unfortunately, NARA did not provide a count of these. This lack is partially filled by the IG report. For instance, the report states that all the documents released in July and October 2017 were released in full.

In addition, the IG report also indicates a publicly available source for further checking.

This source is another list which NARA released in January 2018. The list, which I call NF18, included ALL documents still containing redactions after the 2017 releases. The list, which was also released pursuant to an FOIA request, consisted of an excel spreadsheet with 22,933 rows. For reasons that are unclear to me, many of these rows were duplicates. The total number of unique documents in NF18 was 21,890. ⁴

This list is the basis for the claim on pages 3 and 6 of the IG report that 21,890 ARC documents still contain one or more redactions after the 2017 releases.

NF18 can thus be a tool for testing the IG report statement that the documents released in July and October 2017 were all released in full. Do any of the July/October documents appear on NF18? The answer is no. But is NF18, and the IG, correct in asserting that there is nothing left redacted in these documents?

Well, the documents were posted on the internet. Just check if a July/October document still has redactions. Alas, this is a task that none of the internet commentators on the ARC have yet attempted.

Going further, one can also check the November/December 2017 releases using the same method. For the count of November/December releases not included in NF18, the results I got were as follows. Of a total of 24,626 unique records released in November 2017, 8856 of these do NOT appear in NF18. Of a total of 4217 unique records released in December 2017, 986 do NOT appear in NF18.

These documents are also available for checking to see if they are indeed released in full. No one has bothered checking these either.

This being the case, and now having some spare time, I have attempted this feat for all of the 2017 releases, and will report on the results in a future post.

Meanwhile, adding these figures all together, I get 16,536 documents “released” in 2017 that are not on NF18, meaning that they were released in full.⁵ Returning one more time to the IG report, this is consistent with the claim in the second paragraph of the report that “approximately 16,000” of the 2017 documents were released in full.

These figures are the main contribution of the IG report to our understanding of the current state of the ARC. Since the IG is using NF18 to track releases, we can also do so for the 16 April 2018 release. This will also be the subject of a future post.

This post is a continuation of my April 2018 post on the “old RIF numbers” formerly used in the JFK Assassination Record Collection (ARC) at the National Archives and Records Administration (NARA).

This ID system is no longer used at NARA, but there are still many documents using it in the important on-line version of the ARC at the Mary Ferrell website.

For those who want to get maximum use out of the MF collection, matching up the old IDs with the current IDs can therefore be very, very helpful, as this post explains.

For others, however, this post will probably have little of interest. Caveat lector.

RIFs old and new

The act which created the ARC, the 1992 JFK Assassination Records Act (ARCA), required NARA to “prepare and make available to all Government offices a standard form of identification or finding aid for use with each assassination record subject to review under this Act.”¹ It further required the National Archivist to “ensure that the identification aid program is established in such a manner as to result in the creation of a uniform system of electronic records by Government offices that are compatible with each other.”²

In compliance with this requirement, NARA created the RIF, short for “Reader Information Form”. The RIF has 17 fields for metadata concerning the record it covers, and a unique 13 digit identifying number, the “RIF number”.³ It is usually attached on top of an ARC document as a cover sheet. As a convenient shorthand, I’ll call all these documents “RIF docs.”

The ARCA requirement to create finding aids was limited, however, to federal government records which were not in the possession of NARA when the ARCA went into effect, or which had text redacted and were therefore not open in full for public inspection. The most obvious example of records which do NOT meet these criteria is the documents from the Warren Commission’s investigation of the JFK assassination. These have been in NARA’s possesion since the 1960s, and most of them have been open to the public for over 50 years. This is why most most records from the Warren Commission do not have RIF finding aids, despite being an integral part of the ARC.⁴

Other examples of ARC records without RIFs include documents from Jim Garrison’s investigation of the assassination, which resulted in the trial of New Orleans businessman Clay Shaw for conspiracy to assassinate President Kennedy. None of these documents have RIF forms either. Overall, however, documents without finding aids make up only a small percentage of the total JFK collection; they are far exceeded by documents for which the ARCA required a finding aid.

The Mary Ferrell collection of ARC documents includes a huge number of both RIF docs, such as the CIA 201 file on Oswald, and non-RIF docs, such as records from the Warren Commission and Garrison investigations. The MF website states it has 101,113 documents from the ARC collection available on line.⁵ I originally took this to refer to the total number of documents on the site, but after taking a hard look at documents counts at MF, I believe that the figure of 101,113 documents refers specifically to RIF docs, and does not include any of the thousands of non-RIF docs in the MF collection.

In addition to both RIF and non-RIF docs, as my previous post explained, the MF collection also includes a huge number of documents which use a different form, labeled “NARA Identification Aid.” This form, by my count, has 32 fields, and uses a different form of identifying number: a 35 character string which consists of a date-time string followed by a 6 digit number. (See my previous post for several examples of these.) These forms are only used for CIA documents.

This is not a third set of distinct documents. Instead, it turns out that this form was produced by the CIA when it began processing its records for the ARC, all the way back in 1992. Based on the date-time strings, the forms in the MF collection were created between May 1993 and January 1995. This CIA form was later replaced by NARA’s RIF; the old forms were then removed from documents in the ARC, and the new RIF forms were added in their place.

Why do so many of these forms remain in the MF collection? I believe this is because they were still in use at the time these documents were acquired by groups such as the Assassination Archive Research Centter (AARC), who in turn provided them to MF.

In my previous post I called these older forms “old RIFs” and the id numbers on them “old RIF numbers”. Since the function of the form was the same as the RIF, I will continue to use the name “old RIF” to refer to this earlier type of finding aid.

Old RIF doc count

Getting a count of old RIF docs in the MF collection is challenging; my current count is about 23,300. This is a much larger number than I estimated in my previous post; recall that the MF Foundation’s estimate of their RIF doc total was 101,113, so this potentially increases the size of the MF collection by 20 percent.

These documents are distributed in the MF collection as follows:
HSCA segregated collection: 18,819
Oswald 201: 3653
HSCA microfilm collection: 880

The 23,300 figure includes duplicate numbers; accounting for these, the total number of unique old RIF numbers in the MF collection is about 22,600. There were a fair number of typos in the old RIF numbers which I have corrected to the best of my ability.

Matching old RIFs to new RIFs

As the previous post noted, it is possible to match the old RIFs to the new RIFs. The 35 character old RIF number for these documents was not just dropped. Instead, it was moved to the Comment field of the RIF form, together with the “box number” and “vol/folder number” on the old RIF. I have now found some documentation for this move.

ARC record 104-10331-10342 (dated 27 June 1994) is a discussion of changes NARA requested to the “JFK Database Extract.” Item 6 on the list notes “NARA wants the date time user stamp removed from the ID aid. This will require a modification to the application.” Item 8 notes “NARA would like the box number and folder information prefixed to the comments field. This will require a modification to the application.”

ARC record 104-10331-10343 (dated 28 July 1994) notes that as a result of a 25 July meeting, it was decided to “Move date/time ID number to Comment Section along with box numbers and folder information (#8).”

Based on this, I thought it would be possible to match all of the old RIF docs to the new RIF docs. It turns out, however, that this is not possible. I was able to match only about 21,300 old RIF docs to their new RIF counterparts.

One reason for this is that some of the CIA documents in the ARC have redone the date-time string in the comment field. An example of this is MF docid 95842. This old RIF doc is titled “RPTS ON ACTIVITIES TRAVEL OF THE OSWALDS”, and its id number ‘1993.05.17.17:29:39:000065’ This can be identified as ARC record 104-10001-10015. The current version of this record was released on 26 April 2018, and the comment field is OSW7 : V32 : 20031124-1016188.

How do we know that the these two files represent the same record in the ARC? Merely eyeballing them is NOT adequate. There are far too many duplicates in the ARC to safely claim that two documents have the same RIF number simply because they are identical in appearance. The reason we can say these two documents ARE the same is that there is an earlier release of 104-10001-10015 in the MF collection (docid 49995). In this earlier release, the comment field is OSW7:V32 1993.05.17.17:29:39:000065, exactly matching the ID on the old RIF doc 95842.

Why did the date-time string change in the later release of this record? Apparently, this was one of a set of records that were reprocessed by the CIA in the period 2003-2005,⁶ and these new date-time strings reflect the date of this reprocessing.

Trying to match up old RIF and new RIF records where the date-time string has changed is a mind-boggling task. I have managed to match about 500 of them, and in theory it may be possible to do it for all of them, but at this point I don’t really believe it is worth further time and effort.

New totals for CIA records in the Mary Ferrell Collection

Even without the unmatched records, 21,500 is a lot of records. The MF website’s JFK Database Explorer, which is based on the NARA’s Assassination Collection Reference System (ACRS), gives a total of 85492 CIA records in the ARC. It gives the total number of CIA records in the MF collection as 52314.

The figure of 52314 does not include ANY of the old RIF docs, as far as I can tell. We cannot, of course, simply claim that all 21,500 of the old RIF docs that I was able to match up as new additions to the MF collection. The old RIF docs include a large number of documents that duplicate new RIF documents also on Mary Ferrell, as the examples in the this post and the previous post show.

But, as the excel file linked to in my last post also showed, there are a huge number of old RIF docs that are NOT available as new RIF docs on MF. Thus, the old RIF docs significantly expand the total number of CIA documents available from the MF collection.

At this point I can only provide very rough counts on exactly how many more CIA records the old RIF docs add to the collection. An estimate of 14,000 more records is probably close. This would represent an increase of about 20% from the 52,314 Mary Ferrell has already indexed, to around 66,000. I sent a note to MF on this a while ago, but haven’t heard back from them. Looks like another DIY project. Converting my current database tabulation to an excel sheet is not high among my priorities now, but if you are interested in such a list, leave a comment.

Most of my recent spare time has gone into looking at the ARRB record notices published in the Federal Register (see here for an earlier post on this topic).

I am interested in these notices for several reasons. One important reason is that the record notices include many documents that were omitted from the Assassination Collection Reference System (ACRS), NARA’s database of ARC finding aids. Hopefully these records will be added to the database in the not too distant future, but in the meantime, the Federal Register notices remain the sole online reference for several thousand ARC records.

Another reason for my interest in the ARRB record notices is that they give us a detailed picture of which redactions the ARRB released and which redactions the ARRB sustained in the ARC. This was one of the main tasks of the ARRB set by the 1992 JFK Assassination Record Collection Act (ARCA): to examine the redactions requested by the main executive branch agencies in their classified records, accepting or rejecting these, based on the standards provided by Congress.

In the ARRB’s terminology, accepting a requested redaction was a “postponement,” while rejecting a requested redactions was a “release.” A document with no redactions was described as “open in full,” while a document in which text remained redacted was described as “postponed in part.” In a few cases, entire documents were withheld from public release. These were said to be “postponed in full.”

Every postponement of text in an ARC document was decided by a vote of the ARRB. Releases, on the other hand, sometimes occurred when an agency simply withdrew its request to redact a document and released it in full of its own accord. This of course was usually because the agencies concerned realized that the ARRB would probably vote against retaining their redactions, and rather than defend a losing case, simply dropped it. The ARRB called such releases “consent releases.” According to the ARRB’s Final Report, the majority of releases from the ARC were consent releases.

In order to ensure that the ARRB’s review process was subject to public scrutiny, the ARCA further required the ARRB to publish notices of all postponements of text in the Federal Register.

After going through the Federal Register notices, however, I have not been able to reconcile the postponements in these with the postponements listed in the ACRS. The current version of the ACRS lists over 9700 records as “postponed” either in part or in full. Only 2550 of these records were listed in the Federal Register notices. This seems to contradict my understanding of how the ARRB processed assassination records.

Explanations for some of this, however, can be found in other parts of the ARRB records. An interesting example of this is the microfilm Oswald 201 file. To understand this example, some background on 201 files is necessary. According to the ARRB’s Final Report,

the CIA opens a 201 file on an individual when it has an “operational interest” in that person. The CIA opened its 201 file on Lee Harvey Oswald in December 1960 when it received a request from the Department of State on defectors. After President Kennedy’s assassination, the Oswald 201 file served as a depository for records CIA gathered and created during CIA’s wide-ranging investigation of the assassination. Thus, the file provides the most complete record of the CIA’s inquiry in the months and years immediately following the assassination.¹

As for how Oswald’s 201 file came to be microfilmed, when the House Sub-Committee on Assassinations concluded its investigation of the JFK assassination, it signed a memorandum of agreement with the CIA that

Upon termination of the Committee, all materials provided by CIA and examined by the Committee will be kept and preserved within a segregated and secure area within CIA for at least 30 years, unless the DCI and the House of Representatives agree to a shorter period of time….The decision to microfilm was apparently based on two major considerations, as far as we can determine from our records. First, the integrity of the sequestered records had to be maintained. Second, a number of the files that the Assassination Committee requested were active files, and had to be available to allow people to continue conducting their normal activities within the Agency.²

Following its final release of ARC documents on April 26 2018, NARA noted that

Documents included in the Oswald 201 microfilm were not processed for release or posted since it was determined that the microfilm documents are a duplicate of the original Oswald 201 file that is processed and released. The ARRB evaluated these records and determined that they were duplicate files. NARA conducted our own evaluation, which was completed on February 5, 2018. That independent evaluation agreed with the ARRB’s original assessment.³

The ARRB evaluation process is described in more detail in a memo I found in the electronic records of the ARRB, released last year. The memo is from Robert Skwirot, the ARRB’s chief analyst for CIA records, to Laura Denk, the Executive Director of the ARRB for the last few months of its existence. The full text is as follows:

September 25, 1998
TO: Laura Denk, Executive Director

FROM: Robert J. Skwirot

SUBJECT: Sequestered Collection Microfilm Copy of the Oswald 201 File: Review Board staff procedures to confirm that the microfilm copy matches the original Oswald 201.

When copying to microfilm all the records that had been made available to the HSCA, the CIA transferred the Lee Harvey Oswald 201 file to 13 reels of microfilm. The Review Board staff has made an effort to confirm that all of the records on the microfilm copy of the Oswald 201 file can be found in the original Oswald 201 which was reviewed by the Board in 1995 and 1996.

The Review Board staff chose random samples from a printed copy of the microfilm Oswald 201 and verified that they were in the original Oswald 201 file. This task proved difficult since the two copies of the file were not in the same sequence, possibly due to the mechanics of the microfilming or because the Oswald 201 has been disassembled, reviewed, and reassembled so many times.

Members of the Review Board staff spent approximately five days at different times over the past three years engaged in this meticulous work. The page by page comparison of the hard copies of these files was supplemented by CIA database searches to find a match for those records which proved elusive to the Review Board staff. Review Board staff members were able to physcially match each microflim record examined to the corresponding record in the original Oswald 201.

It is likely that the microfilm of the Oswald 201 is a duplicate of the original. Though the Review Board staff examined less than 10% of the microfilm copy of the Oswald 201, no record they viewed could not be matched to a copy in the original 201. The only way to speak with absolute authority on this subject would be to match each and every record. Our survey convinced the CIA team that viewing every record would not be the most productive use of limited staff resources.

The microfilm reels of the Lee Harvey Oswald 201, as well as the printouts from the microfilm, will be transferred to NARA after September 30, 1998. They will be released in full in 2017.

This memo clarifies Board’s policy on the microfilm 201: having already processed a hard copy of the Oswald 201, the Board decided not to waste time on the microfilm 201 which it had good reason to believe was a duplicate of the materials it had already processed. As a result, release of the microfilm 201 file was postponed until the final release date under the ARCA. This decision was not published in the Federal Register.

[Revised May 29]
The JFK Assassination Records Collection Act (ARCA) is the law that established the massive JFK Assassination Records Collection (ARC) at the National Archives and Records Administration (NARA). As federal law, the record of its enactment is about as public as you can get. Yet the passage of time, and ignorance of the basic law-making process, can confuse even something as transparent and public as this.

A recent post at JFKcountercoup, a blog owned by Bill Kelly, is an example of this (here). According to Kelly:

The law [the ARCA] prescribed that all of the government records on the assassination of President Kennedy be released in full to the public by October 26, 2017, twenty-five years to the day Bush signed it into law. When he signed it however, Bush added a rider to the law that provided the President – and only the President, with the authority to continue withholding certain records beyond the October 26, 2017 date on the grounds of national security, where their release would harm American interests.

It was of course impossible for President Bush to change a bill after it was passed by Congress. A federal law in the United States must first be passed by both Houses of Congress. It then comes to the President, and he can either sign it into law, or refuse to sign it, an act called vetoing. The President cannot rewrite a bill that Congress has passed and then sign it into law.

I learned this as a school child in the United States. I don’t understand how Kelly did not. Where Kelly got this idea is not important, but to actually publish this claim on his blog shows surprising ignorance of how laws are made in the United States Congress. Kelly frequently posts on the ARC and his posts are referenced at websites such as Mary Ferrell, so this error is worth a note.

The real legislative history of the ARCA is clearly described in the Final Report of the Assassination Records Review Board. ¹ Summarizing, the ARCA was introduced in Congress as a joint resolution (H.J.Res 454/S.J.Res 282). After hearings and reports, the Senate passed a revised version of the bill on July 27, 1992. The House passed the Senate version on September 30, and it was signed into law by President Bush on October 26, 1992. The only thing that Bush added to the bill was his signature.

This is a minor book-keeping note on NARA 18 (the excel file that NARA released on 26 April 2018).

I had previously found 20 record number errors in NARA 17 (the excel file that NARA released on 15 December 2017), and posted these in a table here.

I have now gone over NARA 18 in detail, and all of the errors I found in NARA 17 have been corrected as I suggested. (That weird contortion you see me doing is an attempt to pat myself on the back.) In addition to these corrections, there is one other data change I have found in NARA 18: the files for records 124-10184-10083 and 124-10274-10002 have been switched. I will put this change in the errata table today or tomorrow. Why didn’t I catch this? Because the two files were big FBI releases that had previously been withheld in full and were posted without RIF forms, hence no way to find any inconsistencies in their data.

The rest of NARA 18 looks okay so far. As noted in my second post on NARA 18 (here), the excel file has been normalized to one file/one row. This is truly a mercy and will make searches much simpler in the future. Thanks to NARA for fixing what is broken.

This is my second post on the National Archive’s 4-26 release of documents from the JFK Assassination Records Collection (ARC). In this post I will look at the ultimate resolution of the ‘Withheld in full’ (WIF) documents. The summary: The WIF pantry is now bare. Contrary to claims that thousands of documents were withheld in full, only 11 records listed as WIF were released, and only 5 of these were new documents.

This post requires a revision of a term I have been using in previous posts. John Greenewald, owner of The BlackVault website, filed an FOIA request with NARA for a list of records in the ARC that were not yet released in full. In response, NARA provided an excel file listing 19233 documents. I have been calling this list NARA 18, but the 4-26 release of documents now makes this ambiguous. I will therefore switch terminology. From now on, I will use NARA 18 to refer to NARA’s list of documents released on 4-26, and re-christen the list that Greenewald received from NARA as NF18 (available here).

Similarly, in 2016 NARA responded to an FOIA request filed by Michael Raznitsky with another list. I have been calling this list NARA 16, but for consistency, from now on I will call it NF16 (available here).

The new NARA 18

NARA 18 is an excel file posted here

NARA 18 lists all documents and associated metadata released by NARA up to April 26, with links to the pdfs and mp3 files that NARA posted on the same day. The list is cumulative for all releases of ARC documents since July, 2017, so it has 54,637 rows (I always include the header row when counting rows on these excel files). Each row except for the header represents a document/file.

According to NARA, there are 19,045 records/rows in the new release, but clearly something is wrong because there are 19050 rows that are marked as belonging to the 4/26 release. In these 19050 rows, there are five rows that do not have associated files: rows 16800, 16802, 16805, 17047, and 17064.

Perhaps these specific rows are the reason for the excess rows, but there are other possibilities. It is also worth noting that NARA’s announcement of the new release states there are 18731 files associated with the release, but after downloading everything I had only 18726 files, again indicating that there is a problem with 5 files.

As a result of this problem, the new release takes up rows 1-19051 of the excel sheet (1 is the header row).

The remaining 35586 rows in NARA 18 represent ARC releases in 2017. However, the last excel file for the 2017 releases (posted on December 15, 2017, hereafter NARA 17) had 35556 rows. The reason for the extra 30 rows is that NARA 17 was NOT one file per row; instead, it shoehorned 27 FBI files into 12 rows. In addition, 16 of the 17 .wav files in the July 24 release had 1 page pdf files with them, apparently labels for the content of the audio. All of these have now been fixed to one row, one file.

Finally, it seems that the April 26 excel file is missing one record from the July 24, 2017 release: RIF doc 104-10090-10007 104-10086-10154 (posted as docid-32352827.pdf) is missing from the April 26 sheet. The file is still available at NARA, and has an attached RIF sheet, so not a big deal, just remember if you switch between excel files.

There are other differences between data in the December 15 excel file and data in the April 26 excel file, but nothing else that affects the file/record count.

Record status

The biggest change in the NARA 18 excel file format is the way record status is handled. The excel files for the 2017 releases marked record status as ‘In Part’ or ‘In Full’. ‘In Part’ meant that a copy of the document was open for public inspection at NARA’s reading rooms, but with some of the text redacted. ‘In Full’ meant that a copy of the document was not yet available at the reading rooms, and in some cases, even the finding aid for the document was redacted, with various fields in the form marked [Restricted].

The record status cells in the excel files for the 2017 releases were also sometimes blank or had other text that was hard to interpret. These were again due to technical issues at NARA. One of these issues, I am now sure, was that the metadata for the document in NARA’s master document system was not complete, or was inaccurate. This has been a constant problem with the JFK Assassination Collection Reference System (ACRS), the public database for the ARC.

In NARA 18, however, new terms for document status were adopted. Instead of ‘In Part’, the excel file now says ‘Redact’, and instead of ‘In Full’, it now says ‘Withheld’. This terminology was first adopted in NF18, which was released in January of this year. Note that this new terminology was not retroactively applied in the 35586 rows of NARA 18 which describe the records released last year. It is only used for the 19,045 records which were released on April 26.

For the new release, there are 18700 records with ‘Redact’ status, and 11 with ‘Withheld’ status. The other 334 records/rows have miscellaneous strings that, as I said, are due to gaps in the master database or technical glitches.

The last docs ‘Withheld in Full’: from NF18 to NARA 18

Despite the release in 2017 of over 30000 ARC documents, some still claimed that large numbers of ARC documents were being withheld in full. The April 26 release shows that this is not so. This was already clear in NF 18, which listed only 798 records as ‘Withheld’. These records are now accounted for either by posting on-line or by explanations on the webpage for NARA’s ‘JFK Assassination Records Processing Project‘. Here I’ll try to trace how the 798 withheld records in NF18 match up with the accounting given for the 4-26 release.

In doing this, I went to NARA’s database of RIF data, and looked up the metadata that goes with the these records. I have posted an excel file of these 798 records and their metadata here.

The first column in the excel file is my attempt to account for what categories these records fall into. Based on NARA’s Processing Project figures, it is possible to break down these records as follows:

• Materials not releasable under the ARCA, 65% (514 records)
• Microfilm duplicates of other records, 23% (180 records)
• Missing documents, 10% (79 records)
• Unplayable tapes, (10 records)
• previously released material, (4 records)
• releasable material, (9 records)
• Used typewriter ribbons, (2)

Unreleasable material in the ARC consists of documents covered by sections 10 and 11 of the ARCA, which ruled out public release for three types of records: tax records, records deeded as gifts to libraries, and records with material from federal grand jury investigations.

Most of the withheld records in NF18 were tax records for various figures, mostly from Warren Commission files; these include Oswald, his wife Marina, every Oswald employer they could track down, Michael and Ruth Paine, Jack Ruby, members of his family, his roommate George Senator, his attorney Joe Tonahill, and several of Ruby’s friends or business associates. Many fields are ‘restricted’ in these records, so it is hard to say who some of these people are. Based on four fields: agency, originating agency, subjects, and comments, I count 498 records in this category.

Another set of section 10/11 documents are records that were deeded as gifts to libraries, including the Manchester interviews of Jacqueline and Robert Kennedy. Donor restrictions keep these closed. I count 12 of these in NF18.

The last group of section 10/11 documents are records that consist mostly of grand jury materials and are therefore withheld in full. NARA’s Processing Project tells us there were 5 of these, and I can identify 3 from ARRB notices, but the other two are uncertain. One I would guess to be grand jury material is 124-90091-10143, an affidavit from USCOURTS whose subjects are [restricted] and [restricted]. As for the remaining grand jury record, who knows?

I can thus identify 514 unreleasable records in NF18. But NARA’s Processing Project page tells us there are 520, so that leaves 6 unaccounted for.

The next big group of ‘Withheld’ records in NF18 is the microfilm copy of the CIA’s 201 file on Lee Harvey Oswald. These were withheld in full until NARA could verify whether they had extra material in them that was not in the original documents, which were released in full at least 25 years ago. As the Project page says, they finally determined that the documents were 100% duplicates in February of this year. An incredibly boring job, as I have seen after comparing a thousand or so pages myself over the last few months.

There were 180 records in this category in NF18.

In addition to these, there is a group of 79 records that apparently are listed in NARA’s system with RIF metadata, but are either lost or simply bibliographical ghosts. NARA’s Project page lists all of these, and they were all in NF18.

Then there were 10 tapes from the Rockefeller Commission at the Ford Library. The Assassination Records Review Board found these were not playable or retrievable all the way back in 1998. These are again listed on the Project page and were all in NF18.

NF18 also lists four documents as ‘Withheld’ that had been previously released: 104-10535-10000, 104-10535-10001, and 177-10001-10437 were released in 2017. A fourth document, 124-10286-10391, is in the Mary Ferrell collection.

And at the bottom of the list are two typewriter ribbons that were used to type classified materials for HSCA: 180-10142-10055 and 180-10142-10194.

That leaves nine records from NF18 unaccounted for.

Now for the 11 records in NARA 18 marked ‘Withheld’. Two of these are the NF18 records that were previously released in 2017 (104-10535-10000, 104-10535-10001). These should not have appeared on NF 18, but they did, and are released yet again in NARA 18. 10000 had no redactions left in 2017, and the re-release seems to be almost the exact same file. 10001 was one of the most heavily redacted documents in the 2017 releases (perhaps because it is a Mexican government document) and the re-release adds nothing. There is a duplicate copy of 104-10535-10000, so that is 3 of the 11 ‘Withheld’ records.

In addition, one volume of the microfilm copy of Oswald’s 201 file was released (104-10196-10018), despite the fact that NARA announced the microfilm copy was an exact duplicate of the 201 documents previously released. This is an entertaining release: the pdf consists of 312 pages that all say “Image temporarily not available.”

Two CIA records in NF18 were released: 104-10291-10021 and 104-10291-10022. These lengthy records are from the microfilm version of the HSCA’s Segregated CIA collection. They are personnel files with the names of the personnel blacked out. Both feature documents from the CIA’s non-official cover division. I believe that the 4-26 documents are the first time this type of material has ever been released. Note, however, that both of these files are NBR: not believed relevant. This was not CIA’s conclusion; it was the ARRB’s conclusion.

Two HSCA records in NF18 were also released: 180-10120-10010 and 180-10131-10326. 180-10120-10010 is a subpoena HSCA issued to AG Griffin Bell. Oddly, it is for documents related to Martin Luther King, not JFK. No way of knowing why it is in the JFK Collection. 180-10131-10326 is a note by HSCA’s Dan Hardway about David Phillips’ testimony. It turns out there is a duplicate of this note that was not listed in NF18, 180-10110-10015, which is released in NARA 18 as well.

The sole JCS record in NF18 was also released: 202-10002-10134 includes documents originally from the British and Canadian governments relating to Cuban politics in 1963.

Finally, a chargeout sheet for one of the typewriter ribbons was released (180-10142-10055). The other typewriter is accounted for on the Project page, which assures us it is now open if you want to see it. Maybe I will; it is, after all, a sort of monument to archival enthusiasm.

This leaves four records from NF18 unaccounted for: 124-10175-10480, 179-20002-10389, 179-20004-10021, 180-10116-10076. Perhaps some of these are the unreleasable documents unaccounted for, but then again, maybe not.

At the end of the story then, instead of thousands of previously unheard of records filled with state secrets, we get several dozen RIFs with no documents, boxes of duplicate records, unseen tax forms that we knew we were not going to see, 10 broken cassettes and two tape ribbons. Figures.

Mary Ferrell counts

I pay careful attention to Mary Ferrell Foundation posts, since they generally know what’s what, but their record counts for this release (here) have me scratching my head. The totals for the 4-26 release just don’t add up, no matter how I count them. The count for the withheld documents I sort of understand. Everything that was not listed as ‘Redact’ they counted as withheld in full, but this is simply not right. Some of these records were listed as released in part even in the NARA database. Others are listed as released in part in the ARRB record notices.

The only files that I think MIGHT be previously unreleased are the Kennedy Library records (agency number 176-). There are 250 of these with blank status fields. I found an ARRB notice for one of these (176-10031-10069, noticed in 63 FR 53640, at the second to last meeting), but I have not yet found any of the others in the notices or the NARA database. That doesn’t mean they’re not there.

[5/8/2018 postscript] Forgot to mention above a single document that was definitely not previously released, and for which the ARRB did not publish any notice in the Federal Register. This is docid-32627026.pdf, a file which NARA 18 labels “No RIF – 250 pp” This document comes from the PFIAB. Why did it never get an RIF? Beats me.

PFIAB at first told the ARRB that it was “not subject” to the ARCA. The ARRB did not agree, and designated a number of PFIAB documents as assassination records, so that there are FR notices for 18 PFIAB documents. Although the ARRB’s final report leaves the final resolution somewhat up in the air (pg. 155), apparently the ARRB prevailed; all of the PFIAB documents designated in the record notice are now in the MF collection with RIF sheets, and they are quite interesting as policy documents. This last RIF-less document is equally interesting, especially if you are interested in Bay of Pigs material.

So add one more new document to the other 5 in this release, a somewhat fitting coda to the conclusion of a massive quest.