Rabbit's Warren

[Post script added 2018/5/9]

For those not in the know, most (but not all) records in the JFK Assassination Records Collection at NARA have finding aids called Reader Information Forms (RIFs) Each RIF has a unique 13 digit number that identifies the document. In this numbering system, the first three digits represent the agency that provided the record. In my last post I provided a webpage I compiled of all the agency numbers I had found.

After posting the article, I wrote NARA asking if any information on these was available, and was promptly answered by Martha Murphy, who sent me a list of agency numbers, the full titles of the agencies/archives which provided the records, and the record groups (RG) at NARA where further materials from these agencies can be found.

This is a list I have been looking for for a long time, so I have converted it to a table and posted it here.

I have also added to NARA’s list some of the other agency numbers I found earlier. These are marked with an asterisk to indicate which numbers are not from NARA. This version of the list omits extraneous material in the earlier version showing where I got the numbers, since I now have a more authoritative source. I believe there are still at least one or two agency numbers to add, and perhaps a clarification for the confusing agency number 179. Hopefully the complete list will soon appear at NARA. Consider this a sneak preview.

Post script: The agency number list is up at NARA. It is posted here.
Imagine my embarrassment when I found out that it has been up since 2005! (according to the Wayback Machine.) It does not include all of the agency numbers I found, e.g. Eisenhower Presidential Library, Carter Presidential Library, Presidential Foreign Intelligence Advisers Board. To find these, check my table.

It was difficult to come up with a snappy title for this post. Scare quotes around the word missing are a compromise. Otherwise it was something like “RIF forms that were completed but not included in NARA’s on-line database of ARC finding aids.” Well anyway, now you know what this post is really about.

RIFs and the JFK Assassination Records Collection Database

The JFK Assassination Record Collection (ARC) is a documentary collection held by the National Archives and Record Administration (NARA). It was established by the “President John F. Kennedy Assassination Records Collection Act of 1992,” (ARCA) and assembled under the oversight of an independent Federal agency, the Assassination Records Review Board (ARRB).

Records in the Collection came from a wide variety of sources. This post is limited to a discussion of the set of records which the ARCA required to have individual, detailed finding aids, as opposed to the usual practice for finding aids, which are generally done for larger units, such as boxes or folders.

This turned out to be a very large set of records (over 300,000). In addition, the ARCA also required NARA 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.” The intent was to eventually have an electronic database of these identification aids available for the use of both researchers and the general public.

The finding aids which NARA created are called “Reader Information Forms” (RIFs), and the electronic RIF database which the ARCA envisioned is available for on-line searching at NARA’s website. NARA calls this database the “JFK Assassination Collection Reference System” (ACRS).

RIF prefixes

A key feature of every RIF is the record number. Each record in the Collection is assigned a unique 13 digit record number, called a “RIF number” as shorthand. The 13 digits are divided into groups of 3, 5, and 5, separated by dashes. The 3 digit sequence is called an “agency prefix” and indicates the agency which turned over the record to NARA. The first 5 digit sequence is called the “disk number” and represents the number of a floppy disk which NARA provided for agencies to use in creating RIFs. The second 5 digit sequence is called the “control number” and identifies the individual RIF forms on the floppy disk.

As a result of this design, those familiar with the agency prefixes can read off the source of an ARC record by simply looking at the number. One has to be aware, however, that the agency which turned over the record may not be the originating agency. For instance, a record transmitted to NARA by the CIA (prefix 104) might have come from the FBI (prefix 124). In such cases, the RIF number will begin with 104; finding the originating agency will require consulting other data on the RIF. Still, as a quick reference, it is quite useful.

Unfortunately, despite the obvious utility of the system, I have never been able to find a list of which prefixes represent which agencies. As far as I can tell, there is nothing at NARA or even in the ARRB’s electronic records, which were posted at NARA last October.

My latest attempt at my own list of RIF prefixes is here.

I have two different sources for the prefixes and figures in the list. The “ncount” comes from NARA’s ACRS via the JFK Database Explorer at the Mary Ferrell website. The Explorer is a copy of the ACRS made in 2015 and gives some very useful prefix counts here.

The “dcount” comes from my own check of the Federal Register notices that the ARRB published during its review of assassination records over the four years of its existence (see my posts on the ARRB’s FR notices here). The fact that these two sources give different numbers was expected, because of their different nature. What was surprising was that a number of prefixes in the FR notices were NOT in the ACRS.

Prefixes in FR notices but not in ACRS

There were a total of 9 prefixes which occurred in the ARRB’s Federal Register notices, but not in the ACRS. Since prefixes indicate agencies, this means that the ACRS is missing not just individual records, but all records from 9 specific agencies. These include the Secret Service (prefix 154), Immigration and Naturalization Service (136), the National Security Agency (144), the National Security Council (145), the US Army’s Investigative Records Repository (194), the Eisenhower Presidential Library (203), the Carter Presidential Library (207), and the President’s Foreign Intelligence Advisory Board (206). There is also a puzzling record set with the agency number 117, which the ARRB notice describes as “Department of Justice” records (see 63 FR 40096). The categories of DOJ records in the Collection are complex, so it is hard to say what these are without their RIFs.

(I should note that I have already done one post about the “missing” NSA records here.)

Other agency records not in the ACRS

In fact, based on a close reading of the ARRB’s Final Report, I am sure that there are more than 9 agencies whose records are not included in the ACRS. For instance, page 165 of the Final report states that the House Judiciary Committee agree to release “substantial records” from HUAC (the House Un-American Activities Committee) to the ARRB. There is no sign of these records in either the Federal Register notices or in the ACRS. Ditto for records from the Abzug Committee, mentioned on the same page. Page 156 of the Report mentions records from the Department of the Navy; unless these are under ONI (173), these are not listed either.

Individual records not in the ACRS

In addition to whole agencies omitted from the ACRS, there are extensive omissions of records from agencies which do appear there. The most notable case (which I discussed here) is three FBI record sets: 124-10203, 124-10204, and 124-10223. These sets are absent from the ACRS, but over 1400 records from these sets are on line at Mary Ferrell. According to an email I received from Martha Murphy at NARA, the reason these are absent is that three floppy disks of RIFs from the FBI were unreadable when sent over to NARA.

This has not only been an obstacle to researchers, it had an effect on NARA’s recent record releases as well; most of the typos in record numbers were from these sets, and all metadata for these records was omitted from the excel sheets of the released records which NARA posted on-line, meaning that for people who wanted to do things like count how many records from the FBI were released, there were big gaps.

How many records not in the ACRS?

Simply counting the records in the ARRB notices with prefixes that do not appear in the ACRS gives 5866 missing (and remember that my counts for the record notices are only tentative). This is not the whole picture, however. As an example, the notices list 26 records from the Secret Service, yet the ACRS lists zero SS records, truly a baffling omission. In fact, we can be sure that there are many more SS records in the Collection than just 26. Mary Ferrell has a list of Secret Service records (here) numbering over 670, and the control numbers (last five digits) on these record numbers suggest there are hundreds more. The omission of the IRR records is even more startling; almost 5000 of these appear in the FR notices, yet not one is listed in the ACRS. How such extensive omissions came about is hard to understand. Adding up all the omissions discussed above, I think the total number of missing record in the ACRS could easily be 7000-8000. This is an injustice to the hard work of the ARRB and its staff.

The usual 2 cents

The ARC is one of the most important documentary collections in the world today for cold war history. It is also a monument to the idea of transparency in government. Lack of an adequate reference system has obscured the value and significance of the ARC, and has even opened the door to the wholely false claim that there are still “undisclosed records” held in some imaginary vault. NARA has already indicated that a revised version of the ACRS is underway. It should be given the highest priority.

NARA has come through with a new release of 19,045 ARC documents. There are many things to be said about this release, but for now I will simply put up a very important link to NARA’s JFK Assassination Records Processing Project

This page provides important explanations that, unfortunately, will probably not get mentioned at most of the websites that have been posting about the assassination records for the last couple of years. The most interesting to me is the explanation of what happened to the 795 files listed as withheld in NARA’s response to John Greenewald’s FOIA request. According to NARA, 520 of these are not releasable under the ARCA statute. They are either tax returns, protected from public release under section 6103 of the IRS code, deeded as gifts to various libraries (e.g. William Manchester’s notes and interviews for Death of a President), or contained federal grand jury materials under court seal.

NARA’s FOIA response also listed 180 CIA records “withheld.” According to the documents’ metadata，these were all volumes from the CIA’s 201 file for Lee Harvey Oswald. This was odd, because the massive 201 collection was entirely declassified over 25 years ago. I actually realized yesterday what these were (really, I did!), and NARA confirmed it today: these were the microfilm version of the 201. Everyone always assumed that the microfilm version was identical to the original documents, and simply rushed on to other documents, but poor NARA was stuck with the incredibly boring job of comparing the microfilm to the originals, page by page and line by line. According to the project page, they made the official determination that the two were identical on February 5 (Hence the response to Greenewald still listed them). Being identical, NARA did not post them. If you want copies, try Mary Ferrell, but beware–the Mary Ferrell copies are in some cases 25 years old, with many, many redactions that were actually lifted long ago.

Although that is all interesting, the most important news of all comes in the middle of the page: “We will be updating the JFK Assassination Collection database with updated access status and posting the updated database on the web.” This is the most urgent task facing NARA. The sooner it happens, the better.

All in all though, I have to give NARA credit for one thing: they take their statutory responsibilities dead serious.

I have added some new material to my ARRB lists and revised some of the record counts. This post is a short note on what changed.

Meeting list: new transcripts

I have added another set of meeting and hearing transcripts to my list of ARRB meetings. The list originally included links to transcripts online at Mary Ferrell, but these were not complete. I have since found additional transcripts in the ARRB electronic records posted at NARA last October. The transcripts are in this zip, which is a large collection of ARRB administrative records. Look in the zip under the TRANSCRP directory.

These transcripts include the 1996/9/17 Los Angeles hearing, and the 1997/04/02 DC hearing on the Zapruder film. I downloaded all of these transcripts and posted them here at the warren. Just click on the links in the meeting list.

Not all of these transcripts are identical to the MF transcripts. A particularly interesting variation is in the transcript of the 1996/08/06 public meeting. This meeting was held to discuss how the ARRB should handle the segregated CIA files compiled for the HSCA. It turns out that the meeting included a presentation from Barry Harrelson and John Pereira of the CIA’s Historical Review Group. This very informative presentation was cut out of Mary Ferrell’s version of the transcript, as the link shows. What’s up with that, MF?

Records list: revised counts

I have been working quite a bit on the ARRB record notices; the results of this work will appear mostly as revisions of my record notice list, starting with 61 FR 00048. This was the first notice to include consent releases; including these in the record notices is one reason the notices become so long and complicated.

In reviewing the record notices, I have cross checked the 1996 notices with the ARRB’s Fiscal Year 1996 Report, on-line at MF. This report includes two interesting supplements: 1) a summary of the decisions published in notices 14-23 (as numbered in my list), on pages 9-18 of the Report; 2) Appendix B, a 64 page list of all the records noticed.

Both of these cover only the records the Board actually voted on; they do not cover the consent releases. Since the report is devoted to the Board’s activities and specifically to justifying an extension of the Board’s term, this is understandable, but of course this means any questions about the consent releases in the notices will not be answered here. Another limitation is that recissions are not mentioned at all, nor are records reconsidered distinguished from regular determinations. This makes verifying counts tricky sometimes.

In addition, Appendix B has close to 200 duplicate record numbers in it. The majority of these are FBI records, but there are plenty of CIA duplicate numbers too. At this early stage of the ARC, there were clearly problems with making sure that every record had a unique number. This is not surprising, since the finding aids which enforced the unique number requirement were being filled out by a very large group of people, who no doubt often differed in experience and understanding of the records. Strong evidence then of how large a task it was to compile the ARC.

Overall, the 1996 Report is no substitute for the FR record notices in the period it covers. No quick shortcuts for accuracy here (sigh).

This is the third in a series of posts on the Assassination Record Review Board and the notices it published in the Federal Register. This post introduces the 50 record notices the ARRB published over its four years of existence.

In addition to general oversight powers, the JFK Assassination Record Collection Act gave the ARRB two key functions. First, it was charged with finding as many records relating to the JFK assassination as possible. To do this, it was given significant legal powers, especially in dealing with federal agencies and records. Second, it was charged with reviewing material that federal agencies wished to withhold from release, including documents containing law enforcement and national security information, as well as all manner of personal information, ranging from social security numbers to juvenile criminal records.

To ensure that these functions were performed under public scrutiny, the ARCA required the ARRB to publish notices in the Federal Register for all record decisions, from designating raw files as assassination records to individual deletions in each document reviewed.

The Federal Register notices were not intended to provide a complete picture of the Board’s actions. To find that, one must go to the Board’s Final Report, published in 1998 (here) However, if one is interested in the details of the review process, the FR notices are a great resource. And if there were problems with the review process, this of course is one of the first places to look for them.

My list of ARRB record notices is available here.

This list actually comes from two sources. As noted in an earlier post, the Federal Register is now available on-line, including all back-issues from 1936 on, and the list provides links to all of the notices I could find. From the first record notices in June 1995 up until February 1998, the ARRB published complete lists of every record it reviewed, including per-document counts of “text postponements” (explained below).

Beginning in April, 1998, however, the volume of records was so large (notices often covered thousands of records), that the ARRB switched to simply giving summary counts of record categories. Detailed lists were available for those who requested them, and a number of researchers apparently did so. Copies of these list are also available at NARA, but not on-line. In October, 2017, however, as part of a release of ARC records, NARA posted the electronic records of the ARRB on its website (see here) These files include pdfs of ARRB’s lists of individual records it processed.¹ I downloaded these files, so my list includes links to them.

Notice terminology

ARRB record notices are technical documents and use a terminology that ARRB developed. Chapter Four of the ARRB Final Report gives a detailed description of both the process and terminology of the review process that the ARRB developed. Following is a list of the most common review terms:

Designation: When reviewing documents submitted by Federal agencies, the ARRB was charged with deciding whether it was “assassination related.” If it was, the ARRB would “designate” it as an “assassination record,” and the documents were then processed for inclusion in the ARC. There were several notices of such “designations,” but because these were not linked to specific documents in the ARC, I’ll cover these in a separate post.

Determination: When an agency did not want to release information in an ARC document on law enforcement, national security, or privacy grounds, it would mark the information it wished to withhold, and the ARRB would review the document to see whether the agency request met the ARCA criteria for withholding. Withholding is also sometimes called postponement, since the ARCA stated that except for a very limited set of materials, all documents were to be completely released after 25 years. The ARRB’s item by item votes on agency requests to withhold document text are “determinations” and for complex requests, the notices give a count of how many text blocks in a document are “withheld” and “released.” For some documents, these numbers were in the hundreds. In addition to the text determinations, the ARRB also set a date for release of withheld text. Text that was to be withheld for a full 25 years was marked 10/2017; earlier dates were indicated by month and year.

Consent release: This is described in Chapter Four of the ARRB Final report. When documents scheduled for Board review were simply released by agencies without redaction, the Board referred to these as “consent releases.” These were usually listed in the notices as “Additional Releases” with a standard description: “After consultation with appropriate Federal Agencies, the Review Board determined that the following records from [agency name] may now be opened in full.” As the Final Report notes, consent releases actually make up the majority of records processed by the Board.

Reconsideration: This is of the more difficult terms to define; it is not used consistently either in the notices or the Final Report. To provide a more useful accounting of the notices, I will define reconsideration as re-publishing record information in a notice. This generally indicated that the Board voted on the releases and postponements in the record again. Often, but not always, there is some change in these.

Rescission: A notice contains rescissions when it announces that the Board intends to reconsider a specific record. There is no consistent terminology for this; record determinations are said to be “withdrawn”, “suspended”, “rescinded”, and so on. The difference between rescission and reconsideration is that a rescission notice does not specify what determinations have yet been made. This is obviously a temporary status, and as far as I can tell, in every case where a record determination was “rescinded”, a reconsideration notice was eventually published.

Correction: Correction notices varied from correcting typos to noting that a determination was issued for the wrong record. Considering that tens of thousands of records were noticed in the Federal Register, it is a tribute to the Board’s technical competence that there were not more of these.

Duplicate: Duplicate notices were only published in a few cases. As part of the Board’s standard for review, if a record that had not yet been reviewed was discovered to be a copy of another document that had already been reviewed, the Board automatically attached the same releases and postponements. Duplication was a major issue in the ARC, and will be the subject of a future post.

Records “not believed relevant”: These were records that the ARRB believed “truly had no apparent relevance to the assassination” of JFK. As discussed on pg. 47 of the ARRB Final Report, the Board designated very few records as NBR, but when it did, it automatically postponed them in full until 10/2017, removing them from further Board consideration. A number of the ARC records that were not released in full last year belong to this category. This is another subject for a future post.

Records “containing federal grand jury material”: The ARCA specifically granted the ARRB authority to petition courts to unseal federal grand jury records. The notice in 63 FR 23717 states that for several hundred FBI records, “This information is not believed to be relevant to understanding the assassination of President John F. Kennedy, and the Review Board has decided not to petition the relevant court to unseal this information.” This is the only case I’ve found for this. Apparently this means that these records will not be released in full.

Notice counts

My record notice list gives counts of how many records in each of the preceding categories were noticed. These counts are preliminary, and will no doubt change several times over the next couple of weeks. I’ve been working on these notices for almost two months now, and am still bumping into problems, so I can only estimate that I’ll be done sometime in May (maybe). However, I can already say that there are thousands of records in these notices that are not in NARA’s online database of finding aids, and this will be the subject of my next post.

This is the second of several posts I plan to put up discussing the ARRB notices published in the Federal Register. This post covers ARRB meeting notices, of which there were a lot. Federal agencies such as ARRB are required to announce their meetings in advance, regardless of whether they are open to the public or closed. Meeting notices make up the majority of ARRB notices in the Federal Register, and it was a (relatively) simple task to compile a list of all these.

My meeting list is here.

In compiling my list, I checked both the index to the new on-line version of the Federal Register, and Appendix E of the ARRB’s Final Report, which consists of a meeting list. The two are not completely consistent, so I added notes to my list to indicate where they differ. I believe that the Federal Register notices are all accurate; the Appendix is in error when it differs from the Register.

The one exception is the last two meetings listed in the Appendix, on September 28-29, 1998. I have found no notice for either of these in the Register. There was certainly a meeting on September 28, because notices were later published of record determinations made at this meeting. If there really were no notices for these meetings, this would be the only instance that the ARRB omitted these.

The list also includes the six public hearings that the ARRB held, but not the two experts’ conferences, since these were by invitation and therefore not announced in the Federal Register. I’ve attempted a count of the meetings in the list, but this was a bit tricky. There are cases where the ARRB briefly conducted Board business just before or after a hearing. Legally, I suppose these must be counted as meetings to be valid board decisions, but they were not always announced as such. Where the notice stated there would be a meeting and hearing, I counted this; where the notice did not state there would be a meeting in conjunction with the hearing, I did not count this.

Part of the reason for the count was to get some idea of what percentage of the meetings were open to the public and what percentage were closed. Adding in the two unnoticed meetings mentioned above, I count 71 separate meetings: 48 closed meetings, 23 open meetings. Since the Board’s main activity was deciding which of the ARC records should be released immediately and which should be postponed, this ratio is about what I would expect.

In addition to the Federal Register notices, I’ve also added links to materials from Board hearings and meetings which are available on Mary Ferrell. The most complete set of ARRB hearing materials, however, is on Professor John McAdams’ website.

When Congress mandated the creation of the JFK Assassination Record Collection, it wanted to ensure maximum public access to the process. To do this, it decided to impose strict requirements on the Assassination Record Review Board (ARRB), the independent agency it established to oversee the Collection’s creation. One of these requirements was that ARRB had to publish notices of its meetings, decisions, and rule-making in the Federal Register.

The Federal Register is the official ‘newspaper’ of the United States federal government, where all government agencies publish Sunshine Law notices of meetings and rules. These Federal Register notices are thus a permanent public record of the meetings and conduct of the ARRB. The notices, and the massive paperwork preparation for them, take up a good part of the ARRB electronic records that were released by NARA last year, and in turn these electronic records provide a very useful supplement to the Board’s Federal Register notices, as we will see further down the road.

Since the ARRB played the central role in creating the JFK Assassination Record Collection, these notices are an inescapable part of evaluating how the ARC was put together and what is in it. That is why I am going to be posting a fair amount of material on the Federal Register Notices in the next week or so.

The ARRB notices date from 1994 to 1998, so it took some effort to find a source for these. Unfortunately, I did not look hard enough at current resources, and after getting lost in the underbrush at the Office of the Federal Register I went to the Hathitrust collection. Hathitrust actually has large chunks of past issues of the Federal Register. These come mostly from the University of Illinois at Champagne-Urbana; apparently there was at one point a project to scan UIUC’s entire Federal Register holdings. However, the scanned issues are quite incomplete, and the OCR quality was totally inadequate for my purposes.

It turns out that there is now a complete electronic version of the Register on-line at the GPO website. This project was just completed (see here). For 1995 and later, it is down to the document by document level, so I was able to get both PDF and text versions of these notices. For the five notices printed in 1994, one can still download the relevant sections of the Register in whole, and just flip through the pdfs until you find what you need.

Links to, excerpts from, and discussions of ARRB notices in the Federal Register will therefore be posted here in the next week or so.

This is a very technical post. It is intended for people who are seriously interested in finding ARC documents on-line at the Mary Ferrell Foundation website. Everyone else can skip this one.

RIF documents in the ARC

The JFK Assassination Record Collection (ARC) is the great repository of all Federal agency records related to the assassination of President John F. Kennedy, and includes many state and private records as well.

There is a wide range of records and documents in the ARC, but one of the most important and extensive record sets is the records which were transferred to the National Archives and Record Administration (NARA) after the Assassination Record Collection Act (ARCA) was signed into law on October 26, 1992.

The ARCA required that each of these records should have an identification aid attached to it (now usually added as a cover sheet on top of each document). The current version of this aid is called a Reader Identification Form (RIF). As shorthand, I’ll call documents that have an associated RIF “RIF documents.”

One of the main points of the ARCA requirement for an identification aid was to enable the compilation of an electronic database of assassination record data. NARA carried through on this intention, and according to them, the latest version of their on-line public database has data from 319,000+ RIFs.¹

The Mary Ferrell Foundation website (MF) has the largest collection of ARC documents available on-line. Of the 319,000+ RIF docs listed in NARA’s database, MF claims it has 101,113 on line.² This surpasses even NARA’s website, which currently has about 35,000 RIF docs on line.

As this post will show, however, MF’s count is, if anything, an understatement; it has thousands more than that. This was a surprising discovery for me, and I am not sure if MF itself knows just how much it has under-counted its collection. The rest of this post explains where these ‘extra’ documents come from.

Old RIFs and new RIFs

Each RIF provides detailed information for the document it is attached to. Current RIFs consist of 17 data fields plus a unique 13 digit identifying number (RIF number). RIF numbers are very convenient in referring to specific documents and are now almost universally used in research and discussion of the ARC.

A RIF number is divided by dashes into three sets of digits (3-5-5). The first three digits represent the agency that provided the record; this is the agency number. The next 5 digits represent a specific floppy disk which each agency used to record RIF data for the documents it turned over to NARA; NARA calls this the ‘disk number.’ The last 5 digits represent one specific document described in a RIF on that specific disk; NARA calls this the ‘control number.’

When MF says it has 101,000+ documents, it is referring to documents for which these RIF forms have been created. In addition to these RIF documents, however, MF also has a large number of ARC documents which use a different identification form and numbers. I believe these are earlier versions of the RIF form, so I will distinguish the two forms as ‘new RIFs’ and ‘old RIFs’, and the documents they are attached to as old RIF docs and new RIF docs. Below are examples of new and old RIFs from MF:

‘new’ RIF (MF docid 18920)

‘old’ RIF 1 (MF docid 95749)

‘old’ RIF 2 (MF docid 109409)

A careful comparison of these documents will show that despite the different identification aids, they are in fact the same document. It will also suggest WHY there are three different versions of the same document on MF.

A ‘new RIF’ such as the one in MF docid 18920 is titled “JFK Assassination System Information Form”, and divided into two sections: “Agency Information” and “Document Information” with a total of 18 data fields. The record number for our example doc is the 13 digit number 104-10001-10129, formatted as described above.

The ‘old RIF’ examples are identical in form, but include slightly different information. This form is labeled ‘NARA Identification Aid’ and instead of the 18 fields in a new RIF, it has 32 fields, some differing in name and/or content from current RIF fields.

The Old RIF ‘agency’ fields include the agency-no, disk no, and control no. mentioned above. Old RIF example 1 has zeros for the agency data fields, but old RIF example 2 has filled these in, and the agency number, disk number and control number are equal to the 13 digit record number in the new RIF, again telling us that all three of these documents are versions of the same ARC record.

The old RIF ‘document id number’ field gives what I would call an accession number: date, time,and a six digit string (Y.M.D.H:M:S:dddddd); presumably this was the date/time the document was registered in the system.

The two old RIF fields “JFK box number” and “vol/folder” are no longer used in new RIFs, but the data from these fields in our examples has been incorporated into the new RIF comment field, together with the old RIF doc ID number; in the example here, we get “OSW7:V29 1993.05.25.08:47:33:370000”. The rest of the comment field is the same in both the old and new RIFs.

The doc title, agency file number, doc date, rec series, and number of pages are also the same, so that the basic data matches in both the old and new RIFs. The old RIF ‘keyword’ field seems to be the same as the new RIF ‘subjects’ field. The old RIF “curstat” field is the same as the new RIF “current status” field. The contents of “curstat” differ in the two old RIFS. In old RIF 1, curstat is “RIF” (probably = ‘released in full’); in old RIF 2, curstat is SAN ( probably = ‘sanitized’).

Why multiple versions?

Why are there three versions of 104-10001-10129 at MF? It seems that these represent different stages in the document’s release. The print date for old RIF 1 is June 2, 1993. Although it is marked ‘RIF’ (released in full), page 6 is blank, except for a stamp that says “Coordination pending.” This means CIA was checking with the FBI whether it was OK to release the information on this page.

Old RIF 2 (marked ‘sanitized’) does not have a print date, but it has a new stamp on page 3, which says “Reviewed by FBI/JFK Task Force on 12/16/94: Release in Part.” Looking at page 6, the text is restored, but blanks and brackets replace names to protect identities.

The new RIF version of the document is marked “Open,” and looking at page 3, the stamp now reads “Reviewed by FBI/JFK Task Force on 11/9/96: Release in full.” Looking at page 6, the names have all been restored.

Distribution of old RIFs

The “JFK box #” and “vol/folder” fields in the old RIF form are interesting. These suggest that the old RIF was specifically intended for CIA documents, and indeed, as far as I can tell, the old RIF is never used for documents from any other agency. (The example given above is an FBI document collected in the CIA’s 201 file for Oswald.)

There are three main CIA record series in the ARC: the “HSCA Segregated Collection”, the “HSCA Microfilm Collection”, and the Lee Harvey Oswald 201 “personality file.” The old RIF files appear in all three of these series, but apparently in no others. Since these are huge series, I can’t yet give exact figures, but there are thousands of old RIF docs, the majority of them in the HSCA Segregated Collection. ‘Box 1’ (MF setid 1694) of the Segregated Collection has over 1000 documents, and most of them use old RIFs. Box 41 has another 1000 old RIF documents, Boxes 14, 15, and 16 all have 700-800 each, and so on.

The Oswald personality file doesn’t have such a high percentage of old RIF docs in any one volume, but they appear in signficant numbers from vol 27 and up (MF labels these ‘RIF based volumes’), and probably total 2-3000.

The HSCA Microfilm collection has an even lower percentage of old RIF documents, but again they appear in almost every roll, from 1 to 71 (rolls 32 to 42 are blank in MF), so maybe another 1000 there.

Since the old RIF documents use accession date/time in their record numbers, we can also estimate when the old RIFs were produced. The record id number for the vast majority of old RIFs in the Segregated Collection and Oswald 201 Collection are from May to August, 1993. This makes sense, because the ARC was opened to the public at the end of August, 1993 (a date fixed in the ARCA). These RIFs probably represent the government attempt to have as many documents available as possible when the collection opened.

Oddly, the old RIFs in the HSCA Microfilm Collection records were mostly done in 1994. I have found a handful of old RIFs done in January 1995, and after that there are no more old RIFs. The dates in the new RIF comment fields accession strings, however, go up into the 2000s.

The value of old RIFs

What does this all mean to users of the MF documents? It means if you have patience, you can identify which old RIF documents equal which new RIF documents. This is more meaningful than you think.

In particular, since there are so many old RIF documents on MF, I thought it was reasonable to expect that some of these might be ARC documents which were not available on MF in ‘new RIF’ versions.

As an experiment, I looked up the NARA database RIFs for CIA documents with the prefix 104-10001. According to MF’s JFK Data Explorer, there are 177 documents with this prefix (10000-10176), but only 32 of these are online at MF. I was able to look up all 177 RIFs from the NARA database, and used the accession strings in the NARA RIF comment fields to search on MF.

The result: I found 172 of the 177 documents with prefix 104-10001. There were actually 219 documents with matching accession strings, indicating that there are numerous duplicates, but duplicate or not, that is 140 documents more than MF’s own count, meaning 140 documents are available at MF that I had not known were there.

That is worth repeating again. MF doc count: 32. Actual doc count: 172. For those interested, there is an excel file here which gives links to the old rif docs, cross referenced with the modern RIF number for these docs. If this rate holds true for the old RIF documents in general, there should be thousands more CIA documents on MF than their count indicates. This is good news for those who don’t live next door to NARA.

This is my fourth post on NARA 18, maybe I should start numbering them.

NARA 18 is a list of 22,933 records in the JFK Assassination Record Collection that have not yet been released in full. It is the result of an FOIA request from John Greenewald, who runs the website The Black Vault. The list is available as an excel file at NARA’s FOIA Electronic Reading Room (here) or at the FOIA Online site (here)

John Greenewald’s view

After receiving NARA 18, Greenewald compared NARA 18 with NARA 17, the list of ARC documents processed and posted on-line at NARA in 2017. He found that 3082 (actually 3081) records listed in NARA 18 did not appear in NARA 17.

Based on this, Greenewald originally wrote on the Blackvault website that over 3000 documents had not yet been released at all.¹ He later revised this post to concede that the 3082 figure was probably not correct, but he still holds to the view that there is a deliberate, on-going attempt to withhold information from the public by making it difficult to determine what is released and what is not.

Jimmy Falls’ view

Following Greenewald’s revised posting, Jimmy Falls, a writer for the Who What Why website, posted his take (here). His post not only covers the NARA 17 and NARA 18 lists, it also looks at NARA 16, the list of withheld documents which the Who What Why website got from NARA in a 2016 FOIA request.

Falls offers a new version of Greenewald’s comparison of NARA 2017 and 2018: an excel sheet listing 2901 records that are in NARA 2018 and not in NARA 2017. He repeats the claim that “The majority of the files [on his new list] appear not to have been publicly released.”

Falls also compared NARA 16 with NARA 17 and NARA 18, and found that 375 files originally listed on NARA 16 were not released in 2017, nor do they appear on NARA 18. His conclusion: “there appear to be approximately 3,275 files not yet released to the public.”

This post suffers from the same problems that Greenewald’s post had, and is by no means an accurate count.

New and old comparisons

A look at the new list of 2901 records from Falls shows that it is essentially the same set of records that Greenewald described in his post. The only difference is that Falls has removed the duplicate records from his count.

Each ARC document is supposed to have a unique record number, so why there are multiple documents with identical numbers on the list is a puzzle, but putting that puzzle aside for the moment, there are indeed 2901 unique record numbers on NARA 18 that do not appear on NARA 17.²

Since these records do not appear on NARA 17, we know that they were not posted on NARA’s website last year. But this is by no means the same thing as “never publicly released at all.”

All documents in the ARC that are open to the public are available at NARA and have been so since 1998, when the Assassination Record Review Board finished its work assembling the Collection.

Why then does Falls claim that “The majority of the files [on his new list] appear not to have been publicly released”? Did he visit NARA and fail to find them? Did he write to ask for copies and get no response? In fact, he gives no basis for this claim anywhere in his post, a serious problem for his credibility.

Falls’ claim is especially puzzling because it flatly contradicts the clear content of NARA 18.

NARA 18 classifies each document it lists as either ‘redact’ or ‘withheld’. This is surely the current release status of these records, and I am 99% certain that ‘withheld’ means “postponed in full” and “redact” means “released with deletions.” This is the point of NARA 18, to clarify which records still have withheld portions.

Records “postponed in full” have not been released to the public in any form. The ARC has finding-aids for these records, so we know what they are in a general way, but the text of the documents is not available to the public. This is what Greenewald and Falls are talking about when they mention records that are not “publicly released”.

Records “released with deletions” have anything from a single word to whole pages deleted, to protect either national security, law enforcement needs, or personal privacy. It is true that in some of the documents “released with deletions” the majority of the text is gone, but these are a very small percentage; the difference between postponed in full and released with deletions is usually substantial.

So what does NARA 18 say about Falls’ 2901 records? 795 are marked ‘withheld’, i.e. not released to the public in any form. The other 2106 are marked ‘redact’, i.e. released with deletions. Yet Falls says ‘the majority of the files appear not have been publicly released.’ In other words, he doesn’t believe the record status listed on NARA 18 for some undefined portion of the 2106 documents identified as released with deletions. Why? Which ones?

Rechecking the Records

It is of course possible to resolve this question directly: just go to NARA’s College Park facility and ask to view the records. This is what publicly released means. Go to NARA, ask for the record, and they will let you read it. Unfortunately, I have neither the time nor money to do this. (Besides, Falls might not believe me either.) Instead, I have tried a couple of other things.

First, I checked the list of 2901 records against NARA’s online database of finding-aids³ Second, I tried looking for the records on the largest source of ARC documents outside of NARA: the Mary Ferrell Foundation website.⁴

The on-line database at NARA consists of electronic finding-aids for documents in the ARC called Reader Information Forms (RIFs). It does not have RIFs for every document in the collection, but it does have RIFs for all the documents on Falls’ list.⁵

Most RIFs include a current status field. Checking the on-line RIFs shows these are not completely consistent with the information on NARA 18. The on-line RIFs show the current status of 1803 records as ‘released with deletions’, 1075 as ‘postponed in full’, 10 as ‘open’, and leaves the current status field blank for 13.

The main discrepancy is that instead of 795 records ‘withheld’, the on-line database lists 1075 as ‘Postponed in full’, a different of 280. Putting aside this difference for the moment, we are still far from 2901 documents that have never been released to the public.

Perhaps some might still feel free to reject the descriptions in the on-line database; the documents are not on-line to disprove our doubts, perhaps there are massive errors in BOTH the on-line database and NARA 18.

Mary Ferrell is, outside of NARA, the largest source of ARC documents on-line. It does not, however, have all the documents in the ARC. According to an FAQ on Mary Ferrell, their collection consists of approximately one third of the total documents in the ARC.⁶

Of the 2106 records that NARA 18 lists as released with deletions, one can find 1192 on Mary Ferrell. In other words, over half are available in some form at Mary Ferrell, a pretty good rate. This is good evidence that there is no large scale error in NARA 18 or the on-line database. When NARA says these files were released with deletions, they are by and large correct.

I have put the results of checking these two sources in a spreadsheet available here, including the document status as given in NARA’s online database, and links to the files available on Mary Ferrell. Although Mary Ferrell is a subscription service, its ARC documents are available to read on-line for free. Just click on the link.

Discrepancies between NARA 18 and NARA’s on-line database

Returning to the 280 records that NARA 18 says have been released with deletions but the on-line database says are postponed in full, one explanation for the discrepancy is that the on-line database is out of date. In fact, according to NARA it was last updated in 2008.⁷ There is evidence for this at Mary Ferrell, which has at least 6 documents on-line that the on-line database says are withheld in full.⁸ One of these files (124-10286-10391) is also listed as ‘withheld’ on NARA 18. Yet five pages are available at MF. The 795 ‘withheld’ documents should therefore go down to 794.

Of the 35556 records from the JFK Assassination Record Collection (ARC) which were processed by NARA in 2017, one of the more interesting sets was 244 records from the National Security Agency (NSA). Two of these records had been withheld in full; the 2017 release was the first time they became available to the public.¹ The rest had already been released with deletions, many as early as 1997.

The NSA records as historical sources

These records were used extensively by James Bamford in his book on the NSA, Body of Secrets (Doubleday, 2001), which discusses NSA’s role in the U.S. response to JFK’s assassination.² They were also cited by Vincent Bugliosi in his lengthy work on the Kennedy assassination, Reclaiming History (Norton, 2007).³ In addition to works on the JFK assassination, they have also been used in more general discussions of signals intelligence, such as Secrets of Signals Intelligence during the Cold War (Aid and Wiebes, Cass, 2013).⁴

There are a couple of interesting stories in the records that the new releases tell us in more detail than previously available. One newly processed record that caught my attention was the NSA report to the Warren Commission. The Commission had asked NSA to examine Oswald’s possessions for any evidence of cryptographic messages. They found none. The version of this report originally released deleted the names of three of the four NSA employees who conducted the examination, leaving only the name of Meredith Gardner, famous for his work on the Venona decryptions. The new version (144-10001-10289) releases two more names: Ruth Bebb and Carrie Berry. I drew a blank on Bebb, but Berry is now well known, taking up several pages in Code Girls, Liza Mundy’s recent book on women in NSA (Hachette, 2017).

It should be noted that the 2017 releases do not remove all deletions from the records: in many of them, portions of the report title are still deleted (somewhat like the digraphs in CIA sluglines). Paragraph length deletions still remain in the body of most of the records as well, so most of these records show up on NARA 2018, the list of records that still have deletions.

Claims about the NSA records

The NSA records have also recently received attention as a symptom of problems with the continuing release of documents in the ARC. They are mentioned in the Mary Ferrell Foundation’s recent open letter to United States Archivist David Ferriero, which says of them:

We have identified 795 documents that have been released but do not appear at all in the official online NARA database of JFK records. For example, the National Security Agency released 244 JFK records in 2017 that are not listed in the database. Our question: are there other non-disclosed JFK documents from NSA or other agencies that are not in the 2018 listing or scheduled for release?⁵

They are also referenced in a recent post on Jefferson Morley’s website JFKFacts.org, in which Morley claims:

The list of still-secret JFK files is not short. Tens of thousand of files have been released since last October, including more than 200 previously unknown files from the National Security Agency.⁶

The main basis for these claims, in fact the sole basis, seems to be the omission of the NSA documents from NARA’s on-line database of finding-aids for the ARC. On this point, I certainly agree that it would be very useful to have on-line a complete list of all the ARC records with reader information forms (RIFs). I would go further: it would be a very useful project to put copies of all the riffed records themselves on-line. Last year the CIA put its entire CREST document system on-line, and this is an even more extensive collection than the ARC (900,000+ documents, as opposed to 300,000+).

But while I agree with the main suggestions of MFF’s open letter to the Archivist, the underlying theme of the letter is problematic. The idea is that NARA is deliberately, or at least carelessly and culpably, flouting the 1992 JFK ARC Act, and that this is in part a response to pressure from government agencies such as the CIA (or the NSA?) to keep information which is still redacted in the files from being released. This is not credible, and is simply another form of the exaggeration and overstatement that are common in the discussions of the ARC documents on the Internet.

I have already noted one case of “missing documents” in NARA’s on-line database for the ARC.⁷ I will post on this general point again. For now, I’ll consider a few points on the NSA documents in particular. First, has it ever been the case that these documents were unavailable at NARA? Second, did the Assassination Records Review Board, or NARA, fail to disclose the existence of these documents when they were provided by NSA? Third, were these documents generally unavailable, to researchers or the public, prior to their 2017 posting by NARA? The answer to all these questions seems to be no, as one might guess by looking at the first part of this post.

NARA and ARRB handling of NSA records

For the first point, I will just say that I have never heard or read of people who went to NARA before November 2017 looking for the NSA records and didn’t find them. Did NARA ever tell people who asked that they did not have the documents or know of the documents? Again, all comments about the “secret” or “unknown” nature of the NSA records seem to be based only on the fact that they are not listed in NARA’s on-line database of ARC finding-aids. But this is not the only way to learn of the documents.

Second, it is simply not true that the ARRB or NARA failed to disclose the existence of these documents. ARRB’s discussions with NSA and the NSA searches that produced the records are both discussed in detail in the ARRB’s Final Report, as is the number of NSA records and ARRB policy on NSA postponements.⁸

In addition to the Final Report, at the time of the documents’ release, the RIF numbers of many of these postponed documents were published in the Federal Register, as required by the 1992 JFK ARC Act.⁹ In fact, this requirement made it impossible for the Board to conceal records that were found to be assassination-related. After public notice in the Federal Register that the records had been released, as far I know, all such records were made available to any researcher who came to view them at NARA. Was anyone ever told the NSA records were not available? If no, in what way did the ARRB or NARA fail to observe the letter, or the spirit, of the 1992 Act?

As for general availability of the records, the books cited at the beginnning of this post show that interested researchers such as Bamford, Bugliosi, and Aid got access to the documents. Morley’s description of the documents as “previously unknown” is thus a substantial exaggeration. Moreover, Bamford and Bugliosi’s books were widely reviewed, so that through their books even the general public was on notice that assassination-related records from NSA existed at NARA.

NSA handling of the records

In addition to the listings of NSA records from ARRB, and references in works by journalists and researchers, it is also worth pointing out that the NSA website has 373 JFK assassination related documents on-line.¹⁰ These are clearly the same documents that NSA provided to the ARRB in compliance with the JFK ARC Act, but because NSA did not provide the RIF sheets for the documents it posted, and because of duplications in the documents, it is sometimes difficult to identify which of these documents are equivalent to the NSA documents posted by NARA in 2017. I have, however, managed to match up over 200 of the documents on-line at NSA to the documents released by NARA in 2017.¹¹

These documents were posted on the NSA’s website beginning as early as 1999.¹² By 2007, NSA had posted on-line 373 of the 378 assassination related records,¹³ and they remain available on-line today.

Another 2 cents

I strongly agree that a comprehensive database of all documents with finding-aids should be available on-line. After so much time, energy, and money has been devoted to creating the ARC, it is absurd to leave it without this.

As much of the Collection itself as possible should also be on-line. In addition to the documents in the ARC with RIFs, I believe key documents from the Warren Commission should also be on-line. As an example, the 1500+ Commission documents (CDs), which provide the most complete and detailed picture of the basis for the Commission’s Report, are probably more important than many of the relatively minor items that took so much time to produce finding aids for.

As for the unreleased material, I’ll continue to look at the information we have, but in my view there is much less material unreleased than some accounts would have it. I see no point in exaggerating these numbers.