Chapter Seventeen

Reading the Files

Listen to this chapter, read by the author →

What the files are

People imagine the Kennedy files as cables and dossiers. Open one at random and you can get a payroll form. A secretary on the Congressional assassinations committee, appointed the first of June 1977, a grid of little crosses marking her annual leave, and a line in handwriting across the middle: terminated, fifth of August. 10 weeks of somebody’s working life, filed forever under the murder of a president.1

Earlier I said the released files contain private citizens’ Social Security numbers, and that I would not repeat what is on those pages. Here is why that was not a one-off. A completely unrelated document, a different agency, a different year. A payroll form for a committee typist, and her full number is printed on it too. Two for two, from files opened at random. Nobody swept this collection before it went online.2

The name signed at the bottom of that payroll form is worth hearing. Louis Stokes. The congressman who chaired the whole investigation, the man who would stand up in 1979 and tell America the committee believed there had probably been a conspiracy. Here he is in 1977, signing off the termination of one secretary. Every large inquiry is mostly this: a person with a title, approving paperwork.3

Here is an honest number, because nobody covering these releases gives you one. Opening documents at random, roughly one in three has anything in it at all. The rest are cover sheets, index cards on Soviet citizens who never came near Dallas, and staff payroll. That is not a complaint. It is what an archive is, and it is exactly why two and a half thousand documents can sit in public for a year with almost nobody reading them.4

Four months after Dallas, a CIA officer sent a routing sheet to counterintelligence staff about a newspaper clipping. The article had run in an Italian Communist Party weekly and it was titled, roughly, the Kennedy murder and the secret services of the United States. His note flags that a section marked in red refers to rumors that it was the Agency that organized the killing. At the bottom of the form, in handwriting: clipping not retained.5

Nineteen years after Dallas, an allied intelligence service wrote to the CIA asking about one man. Valeriy Kostikov, born 1933, identified as KGB, posted to Mexico City through the sixties and by then serving in Beirut. The letter notes the published claim that he was Oswald’s case officer in Mexico, and asks, on a strictly service to service basis, whether that can be confirmed. It is stamped: do not destroy.6

2,600 documents came out in 2025 and almost none of them are about this plaza. Open them and you find routing slips, folder cover sheets, accountability logs, liaison letters about a KGB officer in Beirut. They are records of intelligence services managing information about themselves. If you are waiting for a file that says what happened on Elm Street, it is not in there, and the absence is itself worth knowing.7

Here is the practical reason nobody has read the new files. They are photographs of paper. No text layer, nothing to search, nothing a computer can index. One document sampled runs to 249 scanned pages. Tens of thousands of pages released into the open, and the only way through them is a person turning pages one at a time. Declassified is not the same as read.8

How paper actually moves

Before we go further into the files I have to explain how information actually moves between agencies, because almost every argument in this subject depends on it and almost nobody sets it out. Start with the thing that surprises people most. A document essentially never travels as itself. What leaves one service and arrives at another is a version, made for the journey, with the parts that would identify where it came from taken out.9

There are two kinds of thing in any of these buildings. Raw material, meaning the actual intercept, the actual tape, the actual informant’s words. And finished product, meaning a report written from that material by an analyst. Partners exchange the second one. They very rarely hand over the first, because the first tells you how they got it. So when a service says it shared everything, that can be entirely true and still mean the raw take never left the building.10

Look at the top and bottom of any of these pages and you will find short words that are not part of the message. They are handling instructions. One says no foreign nationals may see this. One says the originator controls how it is disseminated onward. Another names which compartment you have to be cleared into before the page means anything to you. Those words are not decoration. They are the machinery deciding, in advance, who this piece of paper is allowed to reach.11

Earlier I described the rule that says material you receive from another service is not yours to release. In the files it has a working name, and once you know it you see it everywhere. Third agency. There is a whole folder in the released record titled third agency clearance requests, sixty-one pages, still classified. And there are letters telling a Senate committee that material it asked for is awaiting third agency clearances. That is the rule, wearing its office clothes.12, 13

People picture one enormous vault with everything in it, and a door. That is not what exists. What exists is thousands of small rooms, each with its own list of who may enter, and no room from which you can see all the others. Need to know is not a slogan, it is the wall. Which means a senior officer can answer a question honestly, from inside his room, and be wrong about the building. That has happened in this case more than once.14

Now the part that decides what a researcher can ever find. Bureaus run on registries. A piece of paper is put in a file, the file has a number, the number goes in a central index, and from then on the document is findable by anybody with the index. Take away any step and the paper still exists, still gets read, still gets acted on, and is invisible to every search that will ever be run. It has not been hidden. It has simply never been made findable.15

So when I showed you a cable with not recorded written next to the file number, that was not a curiosity. That is somebody deciding this page will not enter the index. It will move, it will be read, it will be filed somewhere by somebody, and no future search will surface it. I have found that notation more than once in this material. Every time it appears, it means a piece of the record was routed around the only system that would have let you find it.16, 17

Here is the most useful habit I can hand you. On most of these pages there is a box listing who received a copy, and it is often more informative than the message. Sixteen surnames with ticks beside four of them tells you which four people in that building were considered to have an interest. A distribution line that includes a liaison section tells you the paper was being prepared to leave the agency. Read the routing before you read the text.18

And I want to be fair about why things stay closed, because the dull answer is usually right. Overwhelmingly, material is withheld to protect a source or a method: the person who talked, the technique that got it, the fact that a particular room could be listened to at all. Sources outlive presidents. Methods outlive sources. A file kept shut in 2026 is far more likely to be protecting a technique than a killer, and anybody who forgets that will misread everything.19

Hold all of that together and one conclusion follows, and it is the reason this chapter exists. In a system like this, the absence of a document is very weak evidence about the absence of a fact. Something can be known inside a building, discussed by telephone, passed to a friendly service as a sanitized paragraph, and never once written down in a form that any archive will ever surrender. When I say later that the friends produced nothing, understand exactly what that does and does not mean.20

Learning to read them

I opened a cable from the 2025 release and stopped reading. It is marked secret, and one section is marked for the Director only. That section is a key. It takes the cover names the Agency used for its people inside Cuba and sets the real name of each one beside it. Their jobs too. A censor in a state security office. A stevedore on the Havana docks. A postal clerk. In one case, the town where the man lived.21

I have already said I will not read out the pages carrying private citizens’ Social Security numbers. This is a different order of problem. Those were Americans who happened to be somewhere in 1963. These were people informing on a government that is still in power, in a country where their families still live. Sixty-three years is not long enough. I am not naming them, I am not linking the file, and I am not showing the page.22

Here is what that tells you about the release. Four documents I have opened at random now carry things that should not be public. Two Americans’ full Social Security numbers, one private citizen’s home address and life history, and a list of foreign agents by true name. Four out of a few dozen. Nobody went through this collection page by page before it was published. There was not time, and there were two and a half thousand files.23

The rest of the Cuba cable is tradecraft, and it reads exactly like you would hope. Go to the deaddrop. Send me whatever message you find there. Leave the black box. Check the drop every week. It is written in capitals, in cable shorthand, with an X where a period should be. Real spying, it turns out, is a to-do list.24

A different file, August 1962. A New York CIA office reports a new source: an American businessman, born in Havana, twenty-seven years of experience in Cuba, perfect Spanish, in direct contact with an armed exile group. He is not being paid. The memo says he simply wants to be of the best possible service to his government. Fifteen months before Dallas, that is how a source got recruited. A man volunteering because he thought he ought to.25

One sentence in the New York report is worth the whole file. The source will report the ambitions, the operations and the proposed sabotage of the exile group, and the reason given is so that they can either be controlled, encouraged or discouraged. Three options, written flatly, in a government memo. That is the actual American posture toward armed exile groups in 1962, in six words, and nobody was supposed to read it.26

The last line of the New York report is the plan the source brought in. The exile group intended to blow up all the power plants in Cuba. Not a target. All of them. It is recorded in a single sentence at the bottom of a routine contact report, in the same tone as the man’s education and his employment history. That is what fifteen months before Dallas actually looked like on paper.27

Stamped at the foot of the New York report, in 1962: return to CIA, background use only, do not reproduce. It is now a PDF on a government website that anybody in the world can download in about four seconds. I am following the stamp anyway, sixty-three years late, and for a reason the people who applied it never intended, because the page has a living man’s home address on it.28

So here is the rule I have ended up with, and I would rather say it than have you work it out. If a document is about what a government did, I will name it and give you the record number, so you can pull it up and check me. If a page carries a private person’s address, their number, or the name of somebody who spied on a state that still exists, I will tell you what it says and I will not repeat it. Released does not mean fair game.29

Helms

In November 1968 the Director of the CIA wrote to Sir Charles Spry, the head of Australian security intelligence. The letter is marked secret and it thanks him for recommending against declassifying a Warren Commission document. Richard Helms writes that the arguments provided give every reason to keep the document out of the public domain. The head of American intelligence, collecting reasons from abroad to keep a Kennedy file closed.30

The most revealing sentence in the Helms letter is the dullest one. Helms explains that his inquiry a few months earlier had been made in anticipation of further pressure to release Warren Commission papers, and that the pressure had not materialized. Accordingly, there was no present intention to release it. The file stayed shut because the public had stopped asking. That is not a conspiracy. It is just how secrecy actually works.31

For a long time nobody could say what the document actually was, because the letter only gives a number. Commission Document nine seven one. It has since been identified. It concerns anonymous telephone calls made to the American Embassy in Canberra about the assassination, including a claim of Soviet money behind it. So the file an Australian intelligence chief asked Washington to keep shut was a record of tip-offs received in Australia.32

Open two different documents in the 2025 release and you can find the same letter. Same date, same signature, same paragraphs. One a file copy of the other, sitting under two separate record numbers. It is a small thing and it tells you something about the collection: this is not a curated set of revelations. It is an agency’s filing cabinet, emptied out, duplicates and all.33

Consider the tone of it. Dear Charles, to the head of ASIO. Thank you for your letter. With kindest regards. Copies to four departments, drafted by a staff officer, signature recommended by the Deputy Director for Plans. Five years after Dallas, deciding whether the public could see a Kennedy file was ordinary correspondence, handled the way any organization handles paperwork. Secrecy is not dramatic. It is administrative.34

The Helms letter was written in 1968. It was stamped for review in 1976. It came out in March 2025. Fifty-seven years between a man signing a page and anybody outside being allowed to read it, and what it turns out to contain is a polite exchange about keeping a different document closed. That gap, between how long something is withheld and how much it says, is the real shape of this whole archive.35

Sit that beside the call in Cambridge. In England somebody telephoned a newspaper twenty five minutes before the shots. In Australia somebody telephoned the American Embassy about the assassination and what was behind it. Two anonymous callers, two countries, both of them inside the same five nation intelligence arrangement, both generating paper that ended up in American hands. Neither call has ever been traced to anybody.36, 37

And then follow what happened to the Australian one. The calls came in. They became a Commission document with a number. Five years later the head of the service that had handled them wrote to the Director of the CIA arguing that the public should not see it, and the Director agreed and said so in writing. The same allied service that received the tip-offs is the one that later argued for keeping the record of them closed.38, 37

London and Paris

Six months after Dallas, the FBI’s man in London sent a secret cable to Hoover. It relayed information from a listening device inside the Cuban Embassy in London, run by the Security Service, which is MI5 and works inside Britain, and passed onward through the Secret Intelligence Service, which is MI6 and works abroad, to the CIA station. Somebody was recording what Cuban diplomats said to each other about the Kennedy assassination, in Spanish, in a building in London, and the transcript went to Washington.39

In May 1964 an FBI microphone in London picked up a Cuban diplomat and a visitor talking about the assassination. The visitor’s verdict is three words long: a typical F.B.I. job. The diplomat calls it a thing of the imperialists, to kill him in the end, and says it is really disgusting. Six months after Dallas, in a private room in London, the theory was already fully formed, and the FBI was reading it.40

In the same intercepted conversation there is a detail nobody was performing for an audience. Asked whether the man had ever been in Cuba, the diplomat says no. That he applied for a visa at the consulate in Tampico and was refused, and that if he had gone, it would have been in his passport. That is a Cuban official, in private, confirming the man was turned down. The FBI’s own note adds that the name being discussed may have been Oswald. He had in fact applied in Mexico City, not Tampico, which is a small reminder of what raw intelligence is worth before anybody checks it.41

The most honest thing in the London cable is how uncertain it is. The transcript is marked indistinct in places. The name in question is rendered phonetically, as something like Foster, with a question mark. And at the end the FBI writes only that it is possible the person referred to may have been Oswald. This is what raw intelligence actually looks like: a half-heard name in a foreign language, hedged twice, sent on anyway.42

Count the agencies on that single page. British domestic security run the device. British foreign intelligence share it. The CIA station in London receives it. The FBI’s legal attaché cables it to Washington, where it is filed under the assassination case number. Four services in three countries handling one overheard conversation between two men in a room. That is the machinery that surrounded this, and almost none of it was pointed at Dealey Plaza.43

The London cable was written in June 1964. Its cover sheet records a review in 1994, marked for redaction, held back because it needed the approval of another agency. It came out in March 2025. Sixty-one years for one page describing a conversation two Cuban diplomats had about a rumor. Almost everything in this collection is like that: withheld for decades, and then, when you finally read it, ordinary.44

A year after Dallas, somebody sent a letter in Spanish to the American Embassy in Paris. The FBI’s man there passed it up as evidence for the President’s Commission, and then referred the matter to the High General Staff in Madrid, asking Spanish military intelligence to locate the sender, identify him, and interview him for anything he could supply. In 1964, Spain was still Franco’s.45

Read enough of these files and you learn the vocabulary. That Paris memorandum is described as coming from a Controlled American Source, and the document helpfully spells out, in brackets, that this means the CIA. It is a phrase built so a piece of paper can travel between agencies without naming who produced it. Once you can see the euphemism, you can trace how information actually moved.46

The cable ends by promising that the results of the Madrid request will be promptly reported when available. That is where the document stops. Whether the Spanish found the man, whether they interviewed him, whether anything came back at all. None of it is on this page. That is the ordinary experience of reading these files: you get the question, filed and stamped, and almost never the answer.47

Follow the paper. A letter in Spanish arrives at an American embassy in France. The CIA station writes it up. The FBI’s attaché forwards it to Washington for the Warren Commission. The FBI’s Paris office asks Spanish military intelligence to go and find the author. Four countries and three agencies, over one anonymous letter, because in 1964 nobody could afford to ignore anything with Kennedy’s name on it.48

The cover sheet on the Paris file tells its own story. Reviewed in 1997. Status: redact. Held under the section of the JFK Records Act that protects an intelligence source or method. Updated again in 2001. Released in 2025. What was being protected, for six decades, was the fact that the CIA station in Paris passed the FBI a letter and the Bureau asked Madrid about it. The secret was the plumbing.49

Mexico City, up close

Mexico City has been a fog in this book so far, so let me put somebody real in it. When he walked into the Cuban consulate asking for a visa to Havana, the person on the other side of the counter was a Mexican woman in her mid twenties called Silvia Duran, who worked there as a clerk. She dealt with him over more than one visit. She is the closest thing that exists to a witness to what he was actually doing there.50, 51

Her account is not exciting and that is worth something. A man came in wanting to travel to Cuba on his way to the Soviet Union. He produced documents to prove his credentials as a friend of the revolution. She told him he needed a Soviet visa first, which would take weeks. He argued, got agitated, and eventually raised his voice at the consul, who told him to get out. She thought he was a nuisance. That is the whole of it, from the only person who was in the room.51, 52

The day after Dallas she was picked up by the Mexican security police and held. She was interrogated, and her own account of that detention is that it was not gentle. She had not been charged with anything. She was a Mexican citizen, working a clerical job, in her own country, arrested because of a man who had come to her counter seven weeks earlier and been refused a visa.53, 54

Now the part that belongs beside everything I told you about how paper moves. The American station in Mexico City wanted her questioned, and the cable traffic asks the Mexicans to keep her arrest quiet and to ensure that she is not told the interest comes from the United States. So the arrest is theirs, the questions are ours, and the woman in the cell is to be left believing it is a Mexican matter. That is the third agency rule, operating on a human being.54, 52

The station had cameras on those embassies. So after the assassination, when Washington asked for a photograph of Oswald at the Soviet embassy, one was sent. The man in it is heavy set, middle aged, close cropped, and looks nothing whatever like Lee Harvey Oswald. He has been called the Mystery Man ever since and he has never been identified. The photograph is real, it was produced by the station, and the person in it is somebody else.55, 56

†The telephones were tapped too. The Agency’s position afterward was that the tapes had been routinely wiped before the assassination, so no recording of that voice survived. Against that stands the Director of the FBI, on a recorded call to the President the next morning, saying his agents had listened to the tape and the voice was not Oswald’s. Both of those cannot be true. Somebody either destroyed a tape that mattered, or was wrong about having heard it, and it has never been settled which.57, 58, 52

So here is where Mexico City actually stands. A clerk who says he was a nuisance and was refused. A photograph of somebody else. A tape that either never existed or was destroyed, and a Director who said he had heard it. And a foreign government asked to detain a woman without telling her who wanted it. The one thing that would settle any of it is the raw material, and the raw material is the thing that does not survive. That is not a coincidence. It is the design.20

† disputed   ‡ theory   § reconstructed — everything unmarked is established. Superscript numbers key to the sources below. Corrections

Sources for this chapter

58 sources. Every claim above is keyed to one of them. The complete list for the whole book runs to 1,276.

  1. HSCA staff payroll records, personal leave record for 1977 and payroll authorisation form dated 1 August 1977
  2. direct inspection of two unrelated released documents, both carrying unredacted Social Security numbers
  3. HSCA payroll authorisation form, 1 August 1977, signed by the committee chairman
  4. direct sampling of the 2025 release: documents opened and read one at a time
  5. CIA Routing and Record Sheet, 26 March 1964, classified SECRET
  6. Liaison letter, 10 May 1982, reference 6/705, subject Valeriy Vladimirovich Kostikov
  7. 2025 release index, National Archives
  8. direct inspection of released PDFs, no font resources present, image content only
  9. standard intelligence dissemination practice as described in the Church Committee reports and the ARRB final report
  10. Church Committee, Book I, on intelligence production and dissemination
  11. classification and control markings in the released JFK records
  12. RIF 124-10371-10015, HSCA administrative folder, third agency clearance requests
  13. RIF 124-10264-10488, FBI memorandum to the Senate Select Committee
  14. Church Committee findings on compartmentation and internal reporting
  15. FBI records management practice as evidenced in the released files
  16. FBI airtel, Legat Bern to Director, 21 August 1964
  17. compare RIF 124-10150-10104
  18. distribution and routing blocks in the released FBI records
  19. Assassination Records Review Board final report, 1998, on postponement grounds
  20. the author's assessment
  21. CIA cable, page 2, classified SECRET, section marked FOR DIR ONLY
  22. editorial decision on RIF 104-10075-10200
  23. direct inspection of four documents in the 2025 release
  24. CIA cable, page 2, agent instruction text
  25. CIA Office Memorandum, 31 August 1962, classified SECRET
  26. CIA Office Memorandum, 31 August 1962
  27. CIA Office Memorandum, 31 August 1962, final paragraph
  28. CIA Office Memorandum, 31 August 1962, handling stamp
  29. editorial standard applied across this project
  30. Letter from Richard Helms to Sir Charles Spry, 26 November 1968, classified SECRET
  31. Letter from Richard Helms to Sir Charles Spry, 26 November 1968
  32. Letter from Richard Helms to Sir Charles Spry, 26 November 1968, contents of the referenced document not described
  33. direct comparison of two released documents containing the same 26 November 1968 letter
  34. Letter from Richard Helms to Sir Charles Spry, 26 November 1968, with distribution list and drafting notation
  35. Letter from Richard Helms to Sir Charles Spry, 26 November 1968, bearing a 1976 FOIA review stamp
  36. Memorandum from James Angleton, CIA Counterintelligence, 26 November 1963
  37. Warren Commission Document 971
  38. RIF 104-10009-10222, released 18 March 2025, National Archives JFK Collection
  39. FBI AIRTEL, Legat London to Director, 23 June 1964, classified SECRET
  40. FBI AIRTEL, 23 June 1964, quoting a surveillance transcript of 8 May 1964
  41. FBI AIRTEL, 23 June 1964, quoting a surveillance transcript of 8 May 1964; the Bureau's own note that the individual referred to may have been Oswald
  42. FBI AIRTEL, 23 June 1964, transcript marked indistinct, name rendered phonetically, Bureau comment qualified
  43. FBI AIRTEL, 23 June 1964, showing the MI5/MI6 to CIA to FBI chain
  44. RIF cover sheet: date of last review 14 February 1994, status Redact, opening criteria approval of CIA
  45. FBI communication, Legat Paris to Director, 28 October 1964, classified SECRET
  46. FBI communication, 28 October 1964, referring to a Paris Embassy CAS memorandum
  47. FBI communication, 28 October 1964, no follow-up recorded in this document
  48. FBI communication, 28 October 1964, tracing the CAS to Legat to Bureau to HGS chain
  49. RIF cover sheet: original classification Secret, review 24 September 1997, status Redact, restriction JFK Act 6(1)(B), update 23 January 2001
  50. HSCA final report 1979, on the Mexico City visit
  51. statements of Silvia Durán
  52. HSCA final report 1979
  53. HSCA final report 1979, on the detention of Silvia Durán
  54. CIA station cables, Mexico City, November 1963
  55. CIA Mexico City station photographic material, October 1963
  56. HSCA final report 1979, volume III
  57. recorded telephone call, J. Edgar Hoover to President Johnson, 23 November 1963
  58. CIA statements regarding Mexico City audio surveillance