NARA 21, part 5: Redacted CIA records – short

This is my fifth note on NARA 21, the updated version of the JFK ARC database (available here). See the first note in this series for more details on the what, when, why of NARA 21.1The list of notes on this subject is available here.

This note continues the topic I took up in my fourth note: What kind of material is redacted in Collection records and how extensive are these redactions?

This post deals specifically with CIA records. Since CIA records make up 86% of the total number of redacted records, they are naturally worth a note by themselves. However, the reader should bear in mind that what is typical for redactions in CIA records may not be typical for FBI, HSCA, and Church Committee records.2For counts of redacted documents in the ARC by agency, see this note.

I should also emphasize that what I write on CIA redactions today will not necessarily be true tomorrow. As I said before, tracking redactions in the ARC is aiming at a fast moving target. NARA 21 is a snapshot of the JFK ARC as of May 17, 2021. More updates are coming, including more releases of redacted material and more accurate counts of documents with redacted text.3See my recent note on what NARA has to say here. So these figures are bound to change soon.

Redactions in CIA records

To avoid the confusion that plagues this subject, I will first define what I mean by “redacted documents”. These are documents available to the public that have some portion of the text held back. This contrasts with “withheld documents”. These are entire documents that are unavailable to the public.

There are no longer any withheld CIA documents in the ARC.4There is, however, a microfilm duplicate copy of the Lee Harvey Oswald 201 file, already discussed here. All CIA documents in the ARC are available to the public, but a significant number of them are redacted: 11,208, according to the latest update of the JFK collection database.

That number needs an explanation. What is the basis for holding back text in ARC documents? The JFK Act, the law which established the ARC, requires the release of all documents in the collection, but enumerates several narrow exemptions for certain types of documents and information. Some of these exemptions are mandatory and cannot be waived.

At this point, however, most text is held back through an exemption that I call 5(g).5I discussed exemptions in this post. I am working on a page on this subject. NARA 21 lists 11,208 CIA documents where the SOLE basis for redaction is 5(g). Under this exemption, an agency must appeal to the President to continue holding back text, and there are strong limitations on the basis of the appeal. Documents redacted under 5(g) are therefore documents where the concerned agencies feel a real need to hold back information.

How significant is this number, though? Before you can really talk about significance, it is a good idea to know what is redacted and how much is redacted. These are tricky questions, and I have decided to look at them in a somewhat different way. I will first look at how many pages each redacted document has. This metric has several advantages in evaluating how much text is cut out of a document.

First, the fewer pages, the less room for redaction there is. In a document with one page, you just can’t cut that much out. In documents with 200 pages, there is a lot more to cut out.

Second, it is easier to count redactions in documents with just a little text. Looking at the two different ARC versions of John Hart’s “Monster Plot” report,6See below for the link. it was hard to see what was being redacted where; I would not like to say how many more redactions were in version one, compared to version two.

With this in mind, here is a table of number of pages per CIA file as defined above; current status is “redacted” and the basis listed is 5(g) only:

pages record count percent
1 5740 ~51%
2 2143 ~19%
3 947 ~8%
4 538 ~5%
5 243 ~2%
6 213 ~2%
7 112 ~1%
8 106 ~1%
9 100 ~1%
10 74 ~0.66%
more than 10 987 ~9%

Notice that more than 50% of the redacted CIA records in NARA 21 consist of only 1 page. This struck me as odd at first, but it turns out that one page docs make up about 40% of the total CIA docs in the ARC. Why is that? We will see below. The remainder of this note will focus on these documents, which I will call “short docs.” NOTE: “short docs” hereafter means this very specific group of records: CIA one page 5(g) redacted documents.

A little less than 40% of the 5(g) redacted CIA records in NARA 21 are 2-10 pages. I will call these “medium docs”. Some of them are actually compound docs, i.e. they consist of more than one document. Because there are a lot of them, and because they can be complicated, I will not look at them now.

On the other hand, 9% (call it 10%) of the redacted files are more than 10 pages. I will call these “long docs.” Many of these are actually whole files, rather than documents extracted from a bigger record. This can actually make them easier to categorize. I will (eventually) devote a separate note to them.

Checking the records: Redacted CIA documents on-line at NARA

How do we know the metadata in NARA 21 is accurate? Because in 2018, most of the ARC records with redactions were posted on-line at NARA, together with an excel file listing some basic information.7See here for the links to download the excel file and all the documents. This allows us to check the redacted records directly.

The 2017-2018 ARC releases are one of the main topics on this blog, so I’ll say no more about them here.8If you need more on this subject, I suggest you look at the blog sitemap. 2017-2018 release notes are at the top of the sitemap, listed in chronological order.

It turns out that there are indeed a few problems with the metadata in NARA 21, even with something as simple as finding all the 1 page CIA records. This is because there are a few cases where page numbers in NARA 21 are off when you compare them with the CIA records posted at NARA. The most ridiculous case is JFK ARC 104-10534-10205, which NARA 21 lists as one page. This is John Hart’s Monster Plot report, mentioned above.9This is one of the Nosenko NBR documents, see here for a note on this subject. The “Monster Plot” is actually 172 pages long. Even weirder is JFK ARC 104-10439-10115, which NARA 21 lists as 1 page but is actually 98 pages; the NARA document seems to be a totally different document from the one listed in NARA 21. There are about five or six docs like this.

It was after sorting through these problems that I came up with a total of 5740 “short docs”. If anyone is interested in a list of the “short doc” document numbers, just ask in the comment box and I will put it up.

Short doc “typology”

The second part of my approach to looking at redacted material is to look at the type of material redacted. One thing I have looked at closely in the ARC is CIA document types. Understanding document types is especially helpful in understanding the type of information still redacted. Without going into the super messy details of how one defines doc types, here is a rough count of doc types in the 5740 short docs:

# doc type doc count percent
1 cables 2770 ~48%
2 memos 1162 ~20%
3 dispatches 232 ~4%
4 info reports 215 ~3.75%
5 routing sheets 114 ~2%
6 notes 82 ~1.5%
7 journal entries 69 ~1%
8 letters 65 ~1%

Here is the explanation for why there are so many one page CIA docs: they are mostly cables, a telegram form which is designed for short communications, and which is also the CIA’s main method of keeping in touch with stations outside the country (and one or two inside the country as well). Of course there are longer cables, but the average cable is indeed one page long. Note this one doc type accounts for almost half of the short docs.

The next most common doc type is a memo. Memos come in a wild variety: office memos, inter-office memos, memos for the record, memos for the file, memos of understanding, memos of conversations, short blue memos asking “what do we know about this?”, long yellow memos trying to figure out what House Committees know about that, and so on and so on. The figure here lumps all of these together. Many memos are of course longer than one page.

Dispatches are usually packages with interesting things inside: airplane manifests, passenger lists, transcripts of ambassadors’ telephone calls, pictures of people outside the Russian embassy, handwriting to analyze, envelopes to check for secret writing, and so on. Most of the one page dispatches are cover memos for the items in the dispatch case.

An information report is a special form, a variant of a cable that transmits information provided by people who know things; some can be quite long, but most are one to two pages long.

Routing sheets are stuck on top of other documents and show who read what and what they thought or looked into after they read it. Not as fun as you might think. The best comments are those scribbled on cables: “God!” “Huh.” “Wrong!”

Journal entries are from the Office of Legislative Counsel, and give a daily tally of the Office’s multifarious dealings on Capitol Hill. They are included in the collection when the OLC acknowledges meetings, letters, phone calls, etc. from the HSCA, 1976-1979. These are not the memos and documents that flowed out of the meetings, just notes of day to day contacts.

The remaining 20 percent or so of the short docs are a labyrinth of esoteric forms: forms for personnel, forms for agents, forms for reports, forms for investigations, forms to start files, forms to close files, etc. If you ever thought the CIA was the one great exception to bureaucratic agencies, filled with adventurous James Bonds and Ethan Hunts rather than bored paper pushers, you are confused, and the short docs are the cure for your confusion. The short docs include an amazing variety of paper work than must have taken brigades of clerks their entire careers to fill out.

None of these, however, are the short cut we need, the quick and dirty method for counting up the types of information redacted in this tall pile of documents. For that, we have to go back to the cables.

Redactions in one page cables

Note that CIA cables have a set format. There is a header, main body, and footer. Footnotes from the cable secretariat may also appear to provide context and commentary. There are sometimes abbreviated versions of cables in the collection, once kept for short term purposes, but the platonic, ideal cable is set.

For cables that follow the format, there is an originator, a CIA officer, listed on top and three more CIA officers listed at the bottom: the releasing officer, the coordinating officer(s), and the authenticating officer. Most of these people were from what is today called Clandestine Services, and their names were never supposed to be made public.

Enter the redactor, whiteout in hand. Cables are a very rich source of material for redactors, and early versions of these cables were pretty much slashed to ribbons. No longer, though, as we shall see.

The header also includes lists of the stations receiving the cable, either for action or for information, as well as the location sending the message. The CIA does not publicize the locations of its stations and so with the exception of the omnipresent “DIRECTOR” (also abbreviated “DIR”), which represents CIA headquarters, all of these are more fodder for the redactor.

A third problem was created by a specialized type of cable: the identity cable. When CIA transmitted the names of people of interest to them, sometimes friends, sometimes enemies, they used letter codes (IDEN A, IDEN B) in the main cables, which were often sent to more than one station for information purposes. Stations that were actually to work with or against these people, however, received a second cable which provided the real identities and personal info of IDEN A, IDEN B, etc. Such basic info of non-agency people was a third common area where the redactor laid a thick coat of whiteout.

Given this set format and this sensitive information, the ARRB developed a very strong-handed method of dealing with the CIA redactors. The ARRB, mentioned many times on this blog, was the independent federal agency created by the JFK Act to oversee the assembly and release of the JFK collection. The ARRB took a very tough approach to cables in particular, and after carefully listening to CIA arguments, went ahead and released big swathes of information. Faced with truly sensitive materials, however, the ARRB sometimes allowed redactions, but they also developed a special format for cables which included an interesting feature: the use of substitute language.

Substitute language consisted mostly of numbers indicating the type of information elided.10See the ARRB Final Report, p. x for a list of all of these. These numbers are included on many (but not all) of the 5(g) CIA cables which NARA has posted on-line. In this shorthand, [03] stands for a CIA officer, [04] for a CIA asset, [06] for general people, etc. Locations are handled by similar means, with very detailed content: [12] is a CIA “installation” (station) in Africa/Near East, [13] is the Far East, [14] is Northern Europe, etc. These already detailed descriptions are further identified by a number for each distinct location: [14-1], for example, is a still unreleased location somewhere in Scandinavia perhaps.

In addition to these aids, of course, the set format of the cables also informs us that, for example, whatever appears after the phrase “ORIG:” on the top left hand of the cable is the name of a CIA officer.

As a result, we already know a tremendous amount about the cables even without the extensive releases of 2017-2018. Add in this newly released text, and there is surprisingly little information in cables still redacted. The following table gives the total number of redactions per document, followed by how many documents are redacted that many times, and what percentage they make up of the total 2770 “short doc” cables:

redactions count percent
1 1220 ~44%
2 688 ~25%
3 395 ~14%
4 201 ~7%
5 112 ~4%
6 50 ~2%

This tells us that documents with 1-2 redactions make up over two thirds of the short doc cables, and about a third of all the CIA short docs. What are these redactions? Most are the names of CIA officers. Almost none of the CIA officers redacted are field officers, however; [03]s appearing in the body of the cable are few and far between. This is because most of the CIA officer names in the body of the cable were released decades ago. Almost all the names currently redacted are hq officers who performed one of the four roles listed above: originating, authenticating, coordinating, or releasing.

The next most common redaction after this is the location of CIA stations. Many of these were also released decades ago. By now most of the Latin American station names are released, and Western Europe stations as well. Northern Europe still has a number of station names redacted. Africa and Near East are largely redacted and almost all of the Far East stations are redacted. Judging from the very few [13] numbers used, this is probably because almost none of these stations have anything related to the JFK assassination.

Are there short cables with really extensive redactions? There are two sets of cables that do have a large number of “redactions”. One is a type of cable called a book message. A book message is a bulk transmission of one message to many stations. An example is JFK ARC 104-10071-10405. This is a cover sheet directing recipients to put together a description of James Earl Ray, assassin of Martin Luther King Jr. The cable was sent to a large number of Western European stations, but this version of the cable simply blocks out the whole list. A second version of this cable was released as JFK ARC 104-10178-10124. This version does not release any names, but breaks the block down into 29 segments. Based on the format, these are all assuredly CIA stations. If we count the individual segments, rather than the one block, this is the most redacted cable I have yet seen in the short docs. There are a couple of other book messages that also conceal a dozen or more locations.

There is a second type of cable that is also “redacted”. An example is JFK ARC 104-10173-10030. Most of the cable is covered by a ruled sheet, that has written on it the word “denied”. The only part visible is the heading. In the few examples I’ve seen of this, the originator field is marked as redacted, meaning that there is a blank box there covering the text. I suppose this means that the name of the originator may eventually be released. The part covered by the ruled sheet, however, is apparently not available. No box, no comments. If the covered part is released, they will have to first find a version that has text. I don’t know what’s up with these cables. There are about six or seven of them.

There is also one cable that is completely blanked out by a redaction box, except for the cite number. This is JFK ARC 104-10510-10090. Don’t know what’s up with this cable either.

Other than these special examples, what is the most redacted short cable out there? I offer two: JFK ARC 104-10093-10106 and 104-10180-10178, where I count 19 and 18 redactions, respectively. Both cables discuss specific individuals, using very detailed identifying information. The redactions are clearly meant to protect the identity of the individuals discussed.


Based on the short docs I’ve looked at so far, the redacted text remaining in the ARC is NOT extensive. If judged per document, it is in fact small, almost minute. There are a lot of these minutely redacted docs. The short doc cables with 1-2 redactions make up about 17 percent of the total remaining CIA redactions. The purpose of the redactions is clear; it is mainly aimed at protecting the identities of CIA officers involved in sending cables in the early 1960s. It is secondarily aimed at protecting the locations where the cables were sent or received. Most of these cables are NOT clearly assassination related. Some, such as the book message on James Earl Ray, have nothing to do with JFK at all. I predict that documents such as these will continue as 5(g) for at least the near future. Others ARE clearly assassination related. I will discuss these in future notes.

There is still more to be said about these particular types of redactions. First, the redacted information is NOT going to be different for each redaction. The names being protected are not each and every one unique. Instead, many of these names certainly appear multiple times. There must be considerable duplication of the information being protected, specifically the names of officers, and when names are released, it will be a few people whose names appear many times.

Despite the limited number of names, however, I don’t expect a sudden revelation of these in the immediate future. Based on the discussion in the ARRB’s final report, these officers meet specific criteria, for example they retired under cover.

Of course this was true of John Whitten, chief of the CIA’s Western Hemisphere division, as well, yet his name was released almost two decades ago. But in his case the ARRB released his name because of the position Whitten held and his influence on the CIA’s investigation of Oswald. Judging from the context of the documents still redacted, none of the people in these documents held such positions. Lacking this rationale, the redactions covering them will probably be removed as they pass on.

Most of the time we have no way of saying which redactions in which documents are hiding which people’s names. (The situation for locations is much different, as noted above.) There is one case where we can predict this, and as it turns out this case is is also very relevant to the redactions still remaining in the collection.

As I have noted more than once, many duplicate CIA documents appear throughout the collection. That is, a single CIA document often appears under multiple record numbers. Whatever names or locations are hidden in these are of course always the same. And when this happens, there are also many instances where the same document is redacted differently, to our considerable enlightenment. Examples are coming in my next post, so stay tuned. All two of you.