Sixty-One Words, Secret Since 1962. The Man Who Classified Them Died In 1971.
A paragraph about a valve check has been withheld since 1962 through thirty-one documents, each derived from the last. The chain has no root: the 1958 guide that classified it was abolished in 1978, and nobody had ever asked.
Background written by the news staff. Contains no argument or recommendation.
COLLEGE PARK, Md. — Sixty-one words. One paragraph, describing a check performed on a valve on a fuel-handling rig. It has been withheld continuously since 4 September 1962.
We can tell you who signed the marking, which thirty-one documents carried it forward, and what the paragraph says. What has not happened, at any point in sixty-four years, is a decision by any living person that it should still be secret.
Classification creep is a mechanism, not a figure of speech. This is what one looks like from the inside.
What derivative classification is
If a new document quotes or paraphrases material that is already classified, the new document inherits the marking. It is stamped with the words "Derived from" and the source document is named on its face.
That new document then becomes a source for the next one, which is stamped in turn.
At no step does anybody reassess the underlying material. The stamp is an instruction to copy. An officer applying it is not being idle: copying is what the rule requires of them, and Professor Idris Hallowell-Boateng, who studies the practice, points out that an officer who reassessed on their own initiative would be exceeding their authority.
Thirty-one documents
We reconstructed the chain from declassification review sheets, the one-page forms an officer completes whenever a document is examined.
June Halloway filed for them in July 2024, was refused, appealed, and received thirty-one sheets in June 2026 — twenty-three months, which is not this newspaper's record but is close to it.
Each sheet names the document reviewed and the document its marking derives from. Laid end to end they run from a 1962 test record through a 1964 annex to a maintenance handbook, a 1969 training syllabus, a 1974 safety bulletin, a 1981 procedural revision and twenty-five more, to a 2021 records-management inventory whose only classified content is the paragraph itself, quoted in an appendix listing what the department holds.
Twenty-nine of the thirty-one sheets give the sixty-one words as the sole reason for the marking.
The chain has no root
Every derivative marking points at an original one. This chain points at item 9(b) of a 1958 classification guide.
That guide was abolished in 1978.
Its successor, still in force, contains a transitional paragraph instructing officers to maintain existing markings pending review.
Nothing in that sentence says when the review happens. It is now forty-eight years old.
Why the review has never been scheduled
We asked. The answer, in writing, is not evasive. It is a description of a queue.
Reviews are ordered by volume of public request. A file that eleven thousand people ask for is reviewed; a file nobody asks for waits. In sixty-four years, the department says, no member of the public has requested this paragraph. Our own 2024 request was for the review sheets, not the paragraph, so it did not enter the queue either.
The second half of the answer is the part we put twice, to be certain we had it right.
Officers in the review office are assessed on wrongful release. There is no corresponding measure for wrongful withholding.
The department confirmed that in writing without appearing to notice that it was a concession.
The people
The 1962 marking carries the signature of Halvard T. Ostrowski-Nunn, one of nine officers holding original classification authority under the 1958 guide. He retired in 1966 and died in 1971.
Clement Vandersloot ran the department's declassification office until 1994 and is the last officer who could have opened the guide, read item 9(b) and broken the loop.
He is ninety-one. Our correspondent reached him by telephone. He remembers the guide, describes its filing system with some affection, and does not remember the paragraph. There is no reason on earth why he should.
The only thing that moves it
Ms Kessler-Amankwah's reply, printed in full below, defends the queue at length and rather well, and then ends on a fact this story cannot get around.
Our written questions of 2 July were themselves a request. The paragraph has entered the queue, and the review that had not been scheduled in sixty-four years is now scheduled, because a newspaper wrote a letter.
She did not offer that as a defence and we do not print it as one. It is the only mechanism anybody in this story could point at, and it has been available to any member of the public, for a filing fee, the whole time.
What the paragraph says
In 2024 a state archive in Helsinki published a bilingual edition of technical material deposited with it in 1968 under an exchange agreement that has nothing to do with any of the above. Item 41 of that edition is a paragraph of sixty-one words about a valve.
Dr. Aino Virtanen-Salo, who edited the volume, had inherited no chain. She had a document, a translator and a publication schedule. She told us she never considered whether it might be secret somewhere else, and that had she considered it she would have published anyway, because the deposit was unconditional.
The paragraph instructs the operator to check the valve seat for leakage at each shift change, to enter the result in the shift log, and to report any leakage before the rig is next used.
That is the whole of it. It tells you to check a valve.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
We stopped asking for the classified paragraph, which nobody was going to give us, and asked instead for the paperwork generated every time somebody looked at it. Thirty-one review sheets each name the document reviewed and the document its marking derives from, so the chain can be reconstructed link by link and drawn; we then had the drawing checked by a specialist who corrected two of the links. The contents of the paragraph came from a foreign archive that had no chain to inherit.
Who we spoke to
- Thirty-one declassification review sheets, 1962–2021, Obtained under freedom of information after appeal. Filed July 2024; refused; appealed; released June 2026, twenty-three months later 2024–2026 Published in full with this piece. Two sheets carry redactions we did not contest and have marked as such.
- Ruth Kessler-Amankwah, Director of information review at the department. Written questions on 2 July 2026; written answer on 24 July; one follow-up put twice July 2026 Her answer is printed unedited. The passage on wrongful release is the reason this piece exists.
- Dr. Aino Virtanen-Salo, Archivist; edited the 2024 bilingual edition in Helsinki. Interviewed by video call twice; supplied the transfer documentation for the 1968 deposit June 2026 Had never heard of the marking and was mildly annoyed to learn of it.
- Clement Vandersloot, Ran the department's declassification office until 1994. Telephone, twelve minutes, at his home 8 July 2026 Remembers the 1958 guide and its filing system. Does not remember the paragraph.
- Prof. Idris Hallowell-Boateng, Works on derivative classification and marking practice. Interviewed by video call; read our chain reconstruction and corrected two links in it June 2026
What we could not confirm
- That the paragraph printed in Helsinki is the paragraph being withheld. The word count matches, the subject line on the review sheets matches, and the rig type matches. We have not seen the withheld original, we are not permitted to, and this identification is an inference we could be wrong about.
- Whether the chain is thirty-one documents or more. A review sheet is produced when a document is examined. A document carrying the marking that has never been examined leaves no sheet, and would be invisible to the method we used.
- What item 9(b) of the 1958 guide covered. The guide's annex is itself withheld. We did not appeal that refusal, because the appeal we were already running had taken twenty-three months, and we record the decision rather than defend it.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Hollow PostSIXTY-FOUR YEARS ON AND THEY STILL WILL NOT TELL YOU
Ran the withholding as evidence that the paragraph describes something catastrophic.
Had that we did not
The 1962 date and the refusal letter, both accurate.
Left out
That the paragraph has been on open shelves in Helsinki since 2024 in a bilingual edition, and that their own reporter could have read it for the price of a library card.
-
Signal & LedgerReview Backlog Passes 1.4 Million Pages As Queue Rules Hold
Trade coverage of the review office's staffing and throughput. The best reporting on the queue that exists.
Had that we did not
The queueing rule, the staffing figures and the throughput, all of which we have relied on and credited.
Left out
The consequence its own numbers imply: a file nobody has heard of is never requested, is never queued, and is therefore secret permanently rather than for a term.
-
They Buriedthis newspaperSixty-One Words, Secret Since 1962. The Man Who Classified Them Died In 1971.
Reconstructed the whole chain from review sheets and printed the paragraph's contents from an archive that inherited no chain.
Had that we did not
All thirty-one sheets, the department's written answer, and the sentence in it that concedes the asymmetry.
Left out
Our headline turns a signature into a character. Ostrowski-Nunn was one of nine officers with that authority, we cannot show he thought about this paragraph for a minute, and our story says so in the twentieth paragraph while the headline says the opposite in the first four words. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Ruth Kessler-Amankwah, director of information review on 2 July 2026 in writing, with the chain reconstruction attached and twenty-one days to reply. Replied on 24 July 2026. Printed in full and unedited.
I will begin with the concession, because you will print it either way and I would rather it were in my own words.
Nobody has read those sixty-one words for the purpose of deciding whether they should still be withheld since 1962. That is true. It is not defensible and I am not going to defend it.
Now the rest, which your draft treats as an accident of design and which is not.
You describe the asymmetry between wrongful release and wrongful withholding as though we had failed to notice it. We noticed it in 1994, in 2003 and in 2018, and each time we kept it, and here is the reason. A wrongful release cannot be undone. A wrongful withholding can be undone at any time by anybody, including by you, and in this instance it has been, by a newspaper with four reporters and a filing fee. Those two errors are not symmetrical in the world, so the measures that catch them are not symmetrical either. You may think that trade is wrong. I would ask you to notice that it is a trade and not an oversight.
On the queue. You have written it up as an absurdity — that the order of review is set by how many members of the public ask. Consider the alternatives. We could review by date, and spend nine years on 1962 while a 2019 file that eleven thousand people want sits behind it. We could review by our own estimate of public interest, which is to say a government department deciding what the public is interested in, and I invite you to write that story instead. Ordering by request is the only rule we have found that a citizen can operate without our permission. Its failure mode is exactly the one you have found: a document nobody has heard of is never asked for. I accept that failure mode is real. I have not been shown a better rule and neither, in your draft, have you.
Two corrections to your account. Our office did not refuse your 2024 request in order to protect the paragraph; it refused because the review sheets themselves carried derived markings, which is the same defect one layer up, and I would rather you wrote that than the version you have. And the 1978 transitional paragraph was drafted by lawyers to prevent an accidental mass release during a change of policy. It has outlived its purpose by roughly forty years. I have asked for it to be rewritten twice.
Finally. You asked, in writing, on 2 July. That request is now in the queue, which means the review you say has never been scheduled is scheduled, and your article will be out of date. I do not say that to score a point. I say it because it is the only part of this story that is actually a mechanism: somebody asked.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.